yield_prediction.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import sklearn\n",
    "from sklearn.ensemble import RandomForestClassifier, GradientBoostingClassifier\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.model_selection import cross_validate\n",
    "from collections import Counter, defaultdict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import csv\n",
    "from rdkit.Chem import MolToSmiles, MolFromSmiles\n",
    "from tqdm import tqdm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = pd.read_csv('Suzuki_rdkit_artificial_no_solventreagent_no_duplicates_30_70.tsv', sep='\\t')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'properties' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[1;32mf:\\programming_projects\\yield_prediction\\yield_prediction.ipynb Cell 4'\u001b[0m in \u001b[0;36m<cell line: 1>\u001b[1;34m()\u001b[0m\n\u001b[1;32m----> <a href='vscode-notebook-cell:/f%3A/programming_projects/yield_prediction/yield_prediction.ipynb#ch0000003?line=0'>1</a>\u001b[0m \u001b[39mfor\u001b[39;00m col \u001b[39min\u001b[39;00m properties\u001b[39m.\u001b[39mcolumns:\n\u001b[0;32m      <a href='vscode-notebook-cell:/f%3A/programming_projects/yield_prediction/yield_prediction.ipynb#ch0000003?line=1'>2</a>\u001b[0m     \u001b[39mprint\u001b[39m(col)\n",
      "\u001b[1;31mNameError\u001b[0m: name 'properties' is not defined"
     ]
    }
   ],
   "source": [
    "for col in properties.columns:\n",
    "    print(col)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0        11.297365\n",
       "1        10.250014\n",
       "2        11.378095\n",
       "3        11.195533\n",
       "4         9.286395\n",
       "           ...    \n",
       "54292     9.656214\n",
       "54293     9.527969\n",
       "54294    11.417345\n",
       "54295    10.992350\n",
       "54296    11.441508\n",
       "Name: property_C1, Length: 54297, dtype: float64"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data['property_C1']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "y = data['yieldclass'].values\n",
    "properties = data.drop(columns=['yieldclass', 'Component_1', 'Component_2', 'Product', 'property_C1', 'property_C2', 'property_P'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def canonicalize_smiles(smiles):\n",
    "    return MolToSmiles(MolFromSmiles(smiles), canonical=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 75938/75938 [00:16<00:00, 4557.56it/s]\n"
     ]
    }
   ],
   "source": [
    "molecules1 = set(data['Component_1'].unique())\n",
    "molecules2 = set(data['Component_2'].unique())\n",
    "products = set(data['Product'].unique())\n",
    "all_mols = molecules1 | molecules2 | products\n",
    "cleaned_mols = set(canonicalize_smiles(smiles) for smiles in tqdm(all_mols))\n",
    "mol_id = {mol: i for i,mol in enumerate(sorted(cleaned_mols))}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "162891"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(data['Component_1']) + len(data['Component_2']) + len(data['Product'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "75938\n",
      "75874\n"
     ]
    }
   ],
   "source": [
    "print(len(all_mols))\n",
    "print(len(cleaned_mols))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "mol_counts = Counter()\n",
    "mol_counts.update(data['Component_1'])\n",
    "mol_counts.update(data['Component_2'])\n",
    "mol_counts.update(data['Product'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('OB(O)C1=CC=CC=C1', 5693),\n",
       " ('COC1=CC=C(C=C1)B(O)O', 2707),\n",
       " ('CC1=CC=C(C=C1)B(O)O', 1505),\n",
       " ('OB(O)C1=CC=C(F)C=C1', 1179),\n",
       " ('OB(O)C1=CC=C(Cl)C=C1', 805),\n",
       " ('OB(O)C1=CC=C(C=C1)C(F)(F)F', 724),\n",
       " ('CC1=C(C=CC=C1)B(O)O', 709),\n",
       " ('OB(O)C1=CC=CS1', 641),\n",
       " ('OB(O)C1=CSC=C1', 632),\n",
       " ('COC1=CC=CC=C1B(O)O', 598)]"
      ]
     },
     "execution_count": 111,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mol_counts.most_common(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# How do molecules distribute according to yield class?\n",
    "\n",
    "An important question is whether some molecules mainly occur in one of the two yield classes, here we define a measure of how concentrated a molecule is to yield class $1$ as:\n",
    "\n",
    "$$\n",
    "\\frac{N_1}{N_0 + N_1}\n",
    "$$\n",
    "\n",
    "Essentially the mean class of a molecule. If the molecule occur about as much in positive as in negative reactions, this value will be close to $0.5$. If it mainly occurs in the positive class it will be close to $1$, while if it mainly occurs in a negative class it will be close to $0$. \n",
    "\n",
    "We can look how this score is distributed in our dataset. A dataset where there would be no bias due to the molecule would be distributed around 0.5, while a problematic dataset has its distribution concentrated at the values 0 and/or 1."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 54297/54297 [00:26<00:00, 2031.24it/s]\n"
     ]
    }
   ],
   "source": [
    "# First tally what classes the molecules occur in\n",
    "class_mols = defaultdict(lambda: {0: 0, 1:0})\n",
    "for record in tqdm(data.to_dict('records')):\n",
    "    yield_class = record['yieldclass']\n",
    "    for mol_key in ('Component_1', 'Component_2', 'Product'):\n",
    "        smiles = canonicalize_smiles(record[mol_key])\n",
    "        class_mols[smiles][yield_class] += 1\n",
    "\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 152,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "defaultdict(<function __main__.<lambda>()>,\n",
       "            {'O=S(=O)(Oc1ccc2ccccc2c1)C(F)(F)F': {0: 7, 1: 0},\n",
       "             'OB(O)c1ccccc1': {0: 5437, 1: 256},\n",
       "             'c1ccc(-c2ccc3ccccc3c2)cc1': {0: 34, 1: 2},\n",
       "             'Clc1cccc(I)c1': {0: 8, 1: 1},\n",
       "             'Cc1ccc(B(O)O)cc1': {0: 1458, 1: 47},\n",
       "             'Cc1ccc(-c2cccc(Cl)c2)cc1': {0: 1, 1: 0},\n",
       "             'CSc1ccc(Br)cc1C#N': {0: 1, 1: 0},\n",
       "             'OB(O)c1ccc(F)cc1': {0: 1129, 1: 50},\n",
       "             'CSc1ccc(-c2ccc(F)cc2)cc1C#N': {0: 1, 1: 0},\n",
       "             'C[C@H](C(=O)O[C@@H](C(=O)N1CCOCC1)c1ccccc1)c1ccc(I)c(F)c1': {0: 2,\n",
       "              1: 0},\n",
       "             'OB(O)c1cccc(O)c1': {0: 147, 1: 29},\n",
       "             'C[C@H](C(=O)O[C@@H](C(=O)N1CCOCC1)c1ccccc1)c1ccc(-c2cccc(O)c2)c(F)c1': {0: 1,\n",
       "              1: 0},\n",
       "             'CN(C)c1ccc(Br)cc1C(F)(F)F': {0: 2, 1: 0},\n",
       "             'OB(O)c1cccc(OCc2ccccc2F)c1': {0: 3, 1: 0},\n",
       "             'CN(C)c1ccc(-c2cccc(OCc3ccccc3F)c2)cc1C(F)(F)F': {0: 1, 1: 0},\n",
       "             'O=C(O)Cc1cccc(Br)c1': {0: 2, 1: 1},\n",
       "             'O=C(O)Cc1cccc(-c2ccccc2)c1': {0: 1, 1: 0},\n",
       "             'Brc1cncnc1': {0: 63, 1: 3},\n",
       "             'c1ccc(-c2cccnc2)cc1': {0: 26, 1: 1},\n",
       "             'BrCCCOc1ccccc1': {0: 0, 1: 1},\n",
       "             'c1ccc(CCCOc2ccccc2)cc1': {0: 0, 1: 1},\n",
       "             'Cc1ccc(S(=O)(=O)N(C(=O)c2ccccc2)c2ccccc2)cc1': {0: 11, 1: 0},\n",
       "             'CC(=O)c1ccc(B(O)O)cc1': {0: 292, 1: 23},\n",
       "             'CC(=O)c1ccc(C(=O)c2ccccc2)cc1': {0: 9, 1: 0},\n",
       "             'CC(=O)Nc1ccc(F)cc1Br': {0: 2, 1: 0},\n",
       "             'OB(c1ccc(Cl)c(Cl)c1)c1ccc(Cl)c(Cl)c1': {0: 1, 1: 0},\n",
       "             'CC(=O)Nc1ccc(F)cc1-c1ccc(Cl)c(Cl)c1': {0: 1, 1: 0},\n",
       "             'COc1ccc(-c2ncc(I)c(-c3ccccc3)c2C(=O)c2ccccc2)cc1': {0: 5, 1: 0},\n",
       "             'OB(O)c1ccsc1': {0: 590, 1: 42},\n",
       "             'COc1ccc(-c2ncc(-c3ccsc3)c(-c3ccccc3)c2C(=O)c2ccccc2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'O=Cc1ccc(Br)cc1': {0: 105, 1: 1},\n",
       "             'CCCCCCCCB1C2CCCC1CCC2': {0: 23, 1: 0},\n",
       "             'CCCCCCCCc1ccc(OC)cc1': {0: 4, 1: 1},\n",
       "             'C[C@@H](Nc1ncnc2c1c(I)cn2COCC[Si](C)(C)C)c1ccccc1': {0: 7, 1: 0},\n",
       "             'O=[N+]([O-])c1ccc(B(O)O)cc1': {0: 154, 1: 21},\n",
       "             'C[C@@H](Nc1ncnc2c1c(-c1ccc([N+](=O)[O-])cc1)cn2COCC[Si](C)(C)C)c1ccccc1': {0: 1,\n",
       "              1: 0},\n",
       "             'N#Cc1ccc(I)cc1': {0: 16, 1: 1},\n",
       "             'COc1ccc(/C(=C(/B2OC(C)(C)C(C)(C)O2)c2ccc(C(F)(F)F)cc2)c2ccc(C)cc2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'COc1ccc(/C(=C(/c2ccc(C#N)cc2)c2ccc(C(F)(F)F)cc2)c2ccc(C)cc2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'Brc1ccc(-c2ccc(Br)s2)s1': {0: 15, 1: 2},\n",
       "             'CC1(C)OB(c2ccc(N(c3ccccc3)c3ccccc3)cc2)OC1(C)C': {0: 60, 1: 3},\n",
       "             'c1ccc(N(c2ccccc2)c2ccc(-c3ccc(-c4ccc(-c5ccc(N(c6ccccc6)c6ccccc6)cc5)s4)s3)cc2)cc1': {0: 2,\n",
       "              1: 0},\n",
       "             'Fc1ccc(Cl)cc1': {0: 19, 1: 0},\n",
       "             'Fc1ccc(-c2ccccc2)cc1': {0: 32, 1: 0},\n",
       "             'CCOP(=O)(OCC)Oc1ccc2ccccc2c1': {0: 12, 1: 0},\n",
       "             'Cc1cccc(B(O)O)c1': {0: 338, 1: 18},\n",
       "             'Cc1cccc(-c2ccc3ccccc3c2)c1': {0: 3, 1: 0},\n",
       "             'O=[N+]([O-])c1ccc(Br)cc1': {0: 151, 1: 0},\n",
       "             'OB(O)c1ccc(F)cc1F': {0: 143, 1: 11},\n",
       "             'O=[N+]([O-])c1ccc(-c2ccc(F)cc2F)cc1': {0: 2, 1: 0},\n",
       "             'CN(C)c1snc2cc(Br)cnc12': {0: 1, 1: 0},\n",
       "             'COc1ccc(B(O)O)cc1OC': {0: 321, 1: 32},\n",
       "             'COc1ccc(-c2cnc3c(N(C)C)snc3c2)cc1OC': {0: 1, 1: 0},\n",
       "             'COc1ccc2c(c1)c(I)cn2S(=O)(=O)c1ccccc1': {0: 1, 1: 0},\n",
       "             'CN(C)c1ccc(B(O)O)cc1': {0: 171, 1: 26},\n",
       "             'COc1ccc2c(c1)c(-c1ccc(N(C)C)cc1)cn2S(=O)(=O)c1ccccc1': {0: 2,\n",
       "              1: 0},\n",
       "             'COC(=O)/C=C1\\\\NC(=O)C(C)=C1Br': {0: 1, 1: 0},\n",
       "             'C=C[B-](F)(F)F': {0: 151, 1: 8},\n",
       "             'C=CC1=C(C)C(=O)N/C1=C\\\\C(=O)OC': {0: 1, 1: 0},\n",
       "             'Fc1ccc(Br)cc1': {0: 64, 1: 2},\n",
       "             'COC(=O)c1ccc(B(O)O)cc1': {0: 286, 1: 23},\n",
       "             'COC(=O)c1ccc(-c2ccc(F)cc2)cc1': {0: 8, 1: 0},\n",
       "             'CCOC(=O)/C=C(/CNC(=O)OCc1ccccc1)OS(=O)(=O)c1ccc(C)cc1': {0: 6,\n",
       "              1: 0},\n",
       "             'N#Cc1ccc(B(O)O)cc1': {0: 396, 1: 57},\n",
       "             'CCOC(=O)/C=C(/CNC(=O)OCc1ccccc1)c1ccc(C#N)cc1': {0: 1, 1: 0},\n",
       "             'CC/C(=C\\\\I)COCc1ccc(OC)cc1': {0: 1, 1: 0},\n",
       "             'CC/C(=C\\\\[C@@H](C)[C@H](OCOC)[C@H](/C=C(\\\\CC)CO)CC)B1OC(C)(C)C(C)(C)O1': {0: 1,\n",
       "              1: 0},\n",
       "             'CC/C(C=O)=C\\\\[C@H](CC)[C@@H](OCOC)[C@H](C)/C=C(/C=C(\\\\CC)COCc1ccc(OC)cc1)CC': {0: 1,\n",
       "              1: 0},\n",
       "             'CC(=O)Nc1nc2ccc(Br)cc2s1': {0: 9, 1: 2},\n",
       "             'Cc1ccc(B2OC(C)(C)C(C)(C)O2)s1': {0: 9, 1: 0},\n",
       "             'CC(=O)Nc1nc2ccc(-c3ccc(C)s3)cc2s1': {0: 1, 1: 0},\n",
       "             'NC(=O)[C@H]1CC[C@H](CNc2nc(Cl)cc(NCc3ccccc3)n2)CC1': {0: 0,\n",
       "              1: 11},\n",
       "             'OB(O)c1ccc(Cl)cc1': {0: 757, 1: 48},\n",
       "             'NC(=O)[C@H]1CC[C@H](CNc2nc(NCc3ccccc3)cc(-c3ccc(Cl)cc3)n2)CC1': {0: 0,\n",
       "              1: 1},\n",
       "             'CCCCCCCCC1(CCCCCCCC)c2ccccc2-c2ccc(Br)cc21': {0: 2, 1: 0},\n",
       "             'CCCCCCCCC1(CCCCCCCC)c2cc(B3OC(C)(C)C(C)(C)O3)ccc2-c2ccc(B3OC(C)(C)C(C)(C)O3)cc21': {0: 11,\n",
       "              1: 0},\n",
       "             'CCCCCCCCC1(CCCCCCCC)c2ccccc2-c2ccc(-c3ccc4c(c3)C(CCCCCCCC)(CCCCCCCC)c3cc(B5OC(C)(C)C(C)(C)O5)ccc3-4)cc21': {0: 1,\n",
       "              1: 0},\n",
       "             'O=[N+]([O-])c1ccccc1Br': {0: 62, 1: 4},\n",
       "             'OB(O)c1cccc2ccccc12': {0: 534, 1: 29},\n",
       "             'O=[N+]([O-])c1ccccc1-c1cccc2ccccc12': {0: 3, 1: 0},\n",
       "             'Cc1cc(=O)oc2ccc(OS(=O)(=O)C(F)(F)F)cc12': {0: 4, 1: 0},\n",
       "             'O=Cc1cccc(B(O)O)c1': {0: 103, 1: 3},\n",
       "             'Cc1cc(=O)oc2ccc(-c3cccc(C=O)c3)cc12': {0: 1, 1: 0},\n",
       "             'COC(=O)c1cc(C(C)(C)C)cc2c1Oc1c(Br)cc(C(C)(C)C)cc1C2(C)C': {0: 1,\n",
       "              1: 0},\n",
       "             'COc1ccc(B(O)O)cc1C=O': {0: 25, 1: 0},\n",
       "             'COC(=O)c1cc(C(C)(C)C)cc2c1Oc1c(-c3ccc(OC)c(C=O)c3)cc(C(C)(C)C)cc1C2(C)C': {0: 1,\n",
       "              1: 0},\n",
       "             'Cc1cccc(C#N)c1I': {0: 0, 1: 1},\n",
       "             'COc1cc(OC)c(CN2CC(C[B-](F)(F)F)C(=O)N(CC(=O)OC(C)(C)C)CC2=O)c(OC)c1': {0: 0,\n",
       "              1: 5},\n",
       "             'COc1cc(OC)c(CN2CC(Cc3c(C)cccc3C#N)C(=O)N(CC(=O)OC(C)(C)C)CC2=O)c(OC)c1': {0: 0,\n",
       "              1: 1},\n",
       "             'BrC(Br)=C(c1ccccc1)c1ccccc1': {0: 5, 1: 1},\n",
       "             'CC1(C)OB(c2cc3cccccc-3c2)OC1(C)C': {0: 11, 1: 3},\n",
       "             'c1ccc(C(=C(c2cc3cccccc-3c2)c2cc3cccccc-3c2)c2ccccc2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'Nc1ccc(Cl)nc1I': {0: 2, 1: 0},\n",
       "             'OB(O)c1cc2c(F)cccc2[nH]1': {0: 1, 1: 0},\n",
       "             'Nc1ccc(Cl)nc1-c1cc2c(F)cccc2[nH]1': {0: 3, 1: 0},\n",
       "             'COc1ccc(S(=O)(=O)n2ccc3cccc(Br)c32)cc1': {0: 4, 1: 0},\n",
       "             'COc1ccc(S(=O)(=O)n2ccc3cccc(-c4ccccc4)c32)cc1': {0: 1, 1: 0},\n",
       "             'COc1ccc(-c2sc(N3C(=O)c4ccccc4C3=O)c(C(=O)c3ccc(Cl)cc3)c2CBr)cc1': {0: 3,\n",
       "              1: 0},\n",
       "             'COc1ccc(B(O)O)cc1': {0: 2602, 1: 105},\n",
       "             'COc1ccc(Cc2c(-c3ccc(OC)cc3)sc(N3C(=O)c4ccccc4C3=O)c2C(=O)c2ccc(Cl)cc2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'O=C(N1CCCCC1Cc1ccccc1)n1cc(-c2ccc(Br)cc2)nn1': {0: 1, 1: 5},\n",
       "             'OB(O)c1ccc(OC(F)(F)F)cc1': {0: 214, 1: 13},\n",
       "             'O=C(N1CCCCC1Cc1ccccc1)n1cc(-c2ccc(-c3ccc(OC(F)(F)F)cc3)cc2)nn1': {0: 0,\n",
       "              1: 1},\n",
       "             'Ic1c2nc(c(-c3ccccc3)c3ccc(cc4nc(c(-c5ccccc5)c5ccc1[nH]5)C=C4)[nH]3)C=C2': {0: 2,\n",
       "              1: 0},\n",
       "             'OB(O)c1ccc(C(F)(F)F)cc1': {0: 674, 1: 50},\n",
       "             'FC(F)(F)c1ccc(-c2c3nc(c(-c4ccccc4)c4ccc(cc5nc(c(-c6ccccc6)c6ccc2[nH]6)C=C5)[nH]4)C=C3)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'O=C(c1ccccc1)c1ccc(Cl)cc1': {0: 22, 1: 0},\n",
       "             'CC(=O)OC[B-](F)(F)F': {0: 24, 1: 0},\n",
       "             'O=C(c1ccccc1)c1ccc(CO)cc1': {0: 1, 1: 0},\n",
       "             'COc1ccc(CN(Cc2ccc(OC)cc2OC)S(=O)(=O)Oc2cccc(Br)c2)c(OC)c1': {0: 6,\n",
       "              1: 0},\n",
       "             'CC1(C)OB(c2cccc(C(=O)O)c2)OC1(C)C': {0: 12, 1: 3},\n",
       "             'COc1ccc(CN(Cc2ccc(OC)cc2OC)S(=O)(=O)Oc2cccc(-c3cccc(C(=O)O)c3)c2)c(OC)c1': {0: 1,\n",
       "              1: 0},\n",
       "             'C[C@@H]1C(=O)C(C#N)=C[C@]2(c3ccccc3)c3nn(C)c(-c4cccc(Br)c4)c3CC[C@@H]12': {0: 3,\n",
       "              1: 1},\n",
       "             'CS(=O)(=O)NCc1ccc(B(O)O)cc1': {0: 2, 1: 0},\n",
       "             'C[C@@H]1C(=O)C(C#N)=C[C@]2(c3ccccc3)c3nn(C)c(-c4cccc(-c5ccc(CNS(C)(=O)=O)cc5)c4)c3CC[C@@H]12': {0: 1,\n",
       "              1: 0},\n",
       "             'Cc1ccc(S(=O)(=O)N(C(=O)c2ccc(NC(=O)OC(C)(C)C)cc2)c2cc(C(F)(F)F)cc(C(F)(F)F)c2)cc1': {0: 2,\n",
       "              1: 0},\n",
       "             'c1ccc([B-](c2ccccc2)(c2ccccc2)c2ccccc2)cc1': {0: 115, 1: 3},\n",
       "             'CC(C)(C)OC(=O)Nc1ccc(C(=O)c2ccccc2)cc1': {0: 2, 1: 0},\n",
       "             'COCOc1cc(Br)c(Br)cc1OCOC': {0: 3, 1: 0},\n",
       "             'CCCCCCOc1ccc(B(O)O)cc1OCCCCCC': {0: 4, 1: 0},\n",
       "             'CCCCCCOc1ccc(-c2cc(OCOC)c(OCOC)cc2-c2ccc(OCCCCCC)c(OCCCCCC)c2)cc1OCCCCCC': {0: 1,\n",
       "              1: 0},\n",
       "             'Nc1ncc(-c2ccc(CN3CCOCC3)s2)c2scc(Br)c12': {0: 1, 1: 0},\n",
       "             'OB(O)c1cccc2cccnc12': {0: 26, 1: 4},\n",
       "             'Nc1ncc(-c2ccc(CN3CCOCC3)s2)c2scc(-c3cccc4cccnc34)c12': {0: 1,\n",
       "              1: 0},\n",
       "             'CCCCCCn1c2ccc(Br)cc2c2cc([N+](=O)[O-])ccc21': {0: 2, 1: 0},\n",
       "             'CCCCCCOc1ccccc1B1OC(C)(C)C(C)(C)O1': {0: 1, 1: 0},\n",
       "             'CCCCCCOc1ccccc1-c1ccc2c(c1)c1cc([N+](=O)[O-])ccc1n2CCCCCC': {0: 1,\n",
       "              1: 0},\n",
       "             'COc1ccc(CN(C2CCCC2)S(=O)(=O)c2cc(Br)ccc2OC)c(OC)c1': {0: 9,\n",
       "              1: 2},\n",
       "             'CC1(C)OB(c2cnc(N)nc2)OC1(C)C': {0: 33, 1: 31},\n",
       "             'COc1ccc(CN(C2CCCC2)S(=O)(=O)c2cc(-c3cnc(N)nc3)ccc2OC)c(OC)c1': {0: 0,\n",
       "              1: 1},\n",
       "             'Cc1ccc(S(=O)(=O)n2ccc3c(-c4cc(Br)ccc4OCC4CC4)cn(C)c(=O)c32)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'OB(O)c1cccnc1': {0: 439, 1: 122},\n",
       "             'Cn1cc(-c2cc(-c3cccnc3)ccc2OCC2CC2)c2cc[nH]c2c1=O': {0: 1, 1: 0},\n",
       "             'N#Cc1cccc(CNc2nc(C(F)(F)F)nc3ncc(Br)cc23)c1': {0: 1, 1: 0},\n",
       "             'CC1(C)OB(c2cnn(C3CCNCC3)c2)OC1(C)C': {0: 3, 1: 0},\n",
       "             'N#Cc1cccc(CNc2nc(C(F)(F)F)nc3ncc(-c4cnn(C5CCNCC5)c4)cc23)c1': {0: 1,\n",
       "              1: 0},\n",
       "             'N#Cc1ccccc1Br': {0: 55, 1: 3},\n",
       "             'N#Cc1ccccc1-c1ccccc1': {0: 9, 1: 1},\n",
       "             'C=CCCN(C=C(Cl)Cl)S(=O)(=O)c1ccc(C)cc1': {0: 3, 1: 0},\n",
       "             'COc1ccccc1B(O)O': {0: 557, 1: 41},\n",
       "             'C=CCCN(/C=C(\\\\Cl)c1ccccc1OC)S(=O)(=O)c1ccc(C)cc1': {0: 1, 1: 0},\n",
       "             'Brc1cc2[nH]c3cc(Br)c(Br)cc3c2cc1Br': {0: 1, 1: 0},\n",
       "             'Fc1ccc(-c2cc3[nH]c4cc(-c5ccc(F)cc5)c(-c5ccc(F)cc5)cc4c3cc2-c2ccc(F)cc2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'COc1cccc(-c2cc(Br)cs2)c1': {0: 5, 1: 0},\n",
       "             'COc1ccc(-c2csc(-c3cccc(OC)c3)c2)cc1': {0: 1, 1: 0},\n",
       "             'OCc1ccccc1Cl': {0: 14, 1: 0},\n",
       "             'CC(=O)c1ccc(-c2ccccc2CO)cc1': {0: 2, 1: 0},\n",
       "             'CC1(C)Cc2nc1cc1cc(C#Cc3ccccc3)c(cc3nc(cc4ccc([nH]4)c2Br)C=C3)[nH]1': {0: 1,\n",
       "              1: 0},\n",
       "             'CC1(C)OB(c2ccncc2)OC1(C)C': {0: 45, 1: 20},\n",
       "             'CC1(C)Cc2nc1cc1cc(C#Cc3ccccc3)c(cc3nc(cc4ccc([nH]4)c2-c2ccncc2)C=C3)[nH]1': {0: 1,\n",
       "              1: 0},\n",
       "             'COc1ccc2ccccc2c1Br': {0: 36, 1: 1},\n",
       "             'OB(O)c1ccccc1-c1ccccc1': {0: 78, 1: 5},\n",
       "             'COc1ccc2ccccc2c1-c1ccccc1-c1ccccc1': {0: 2, 1: 0},\n",
       "             'N#Cc1ccc(Br)cc1': {0: 175, 1: 3},\n",
       "             'CC(=O)c1ccc(-c2ccc(C#N)cc2)cc1': {0: 5, 1: 0},\n",
       "             'Cc1ccc(S(=O)(=O)n2ccc3c(Cl)ncnc32)cc1': {0: 1, 1: 1},\n",
       "             'CCc1cc(B2OC(C)(C)C(C)(C)O2)ccc1CN1C(=O)c2ccccc2C1=O': {0: 0,\n",
       "              1: 1},\n",
       "             'CCc1cc(-c2ncnc3c2ccn3S(=O)(=O)c2ccc(C)cc2)ccc1CN1C(=O)c2ccccc2C1=O': {0: 0,\n",
       "              1: 1},\n",
       "             'Ic1c(OCc2ccccc2)nc2ccccc2c1OCc1ccccc1': {0: 7, 1: 0},\n",
       "             'COc1ccc(-c2c(OCc3ccccc3)nc3ccccc3c2OCc2ccccc2)cc1': {0: 1, 1: 0},\n",
       "             'Clc1nccc2ccccc12': {0: 39, 1: 2},\n",
       "             'c1ccc(-c2nccc3ccccc23)cc1': {0: 2, 1: 0},\n",
       "             'Cc1ccc(S(=O)(=O)Oc2ncc(-c3ccccc3)cc2-c2ccccc2)cc1': {0: 2, 1: 0},\n",
       "             'COc1ccc(-c2ncc(-c3ccccc3)cc2-c2ccccc2)cc1': {0: 1, 1: 0},\n",
       "             'Clc1ncnc2ccsc12': {0: 3, 1: 0},\n",
       "             'OB(O)c1cc2ccccc2s1': {0: 163, 1: 21},\n",
       "             'c1ccc2sc(-c3ncnc4ccsc34)cc2c1': {0: 1, 1: 0},\n",
       "             'Cc1ccc(C(=O)OC[C@H]2O[C@@H](c3ccc(Br)cc3)C[C@@H]2OC(=O)c2ccc(C)cc2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'OB(O)c1ccc2ccccc2c1': {0: 432, 1: 18},\n",
       "             'Cc1ccc(C(=O)OC[C@H]2O[C@@H](c3ccc(-c4ccc5ccccc5c4)cc3)C[C@@H]2OC(=O)c2ccc(C)cc2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'O=c1c(Br)nsnc1-c1ccccc1': {0: 4, 1: 0},\n",
       "             'COc1ccc(-c2nsnc(-c3ccccc3)c2=O)cc1': {0: 2, 1: 0},\n",
       "             'Ic1c(-c2ccccc2)n2nc3c(c2c2ccccc12)CCCC3': {0: 1, 1: 0},\n",
       "             'COc1ccc(-c2c(-c3ccccc3)n3nc4c(c3c3ccccc23)CCCC4)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'Cc1c(-c2ccc(N3CCCC(O)C3)nc2)sc2c(N3CCOCC3)nc(Cl)nc12': {0: 1,\n",
       "              1: 0},\n",
       "             'Cc1c(-c2ccc(N3CCCC(O)C3)nc2)sc2c(N3CCOCC3)nc(-c3cnc(N)nc3)nc12': {0: 1,\n",
       "              1: 0},\n",
       "             'COC(=O)c1c(OS(=O)(=O)C(F)(F)F)c(OS(=O)(=O)C(F)(F)F)c(C(=O)OC)n1CCc1ccc(OC)cc1': {0: 3,\n",
       "              1: 0},\n",
       "             'COc1cc(B(O)O)cc(OC)c1OC': {0: 158, 1: 9},\n",
       "             'COC(=O)c1c(-c2cc(OC)c(OC)c(OC)c2)c(-c2cc(OC)c(OC)c(OC)c2)c(C(=O)OC)n1CCc1ccc(OC)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'Brc1cccs1': {0: 104, 1: 4},\n",
       "             'CC1(C)OB(c2cc3c([Si](C)(C)C)cc4c5cc(B6OC(C)(C)C(C)(C)O6)cc6c([Si](C)(C)C)cc(c(c2)c34)c65)OC1(C)C': {0: 0,\n",
       "              1: 1},\n",
       "             'C[Si](C)(C)c1cc2c3cc(-c4cccs4)cc4c([Si](C)(C)C)cc(c5cc(-c6cccs6)cc1c52)c43': {0: 0,\n",
       "              1: 1},\n",
       "             'C/C(=C/I)CO': {0: 1, 1: 0},\n",
       "             'COc1ccc(Nc2ccccc2B2OC(C)(C)C(C)(C)O2)cc1': {0: 2, 1: 0},\n",
       "             'COc1ccc(Nc2ccccc2/C=C(/C)CO)cc1': {0: 1, 1: 0},\n",
       "             'COc1cc(-c2[nH]c3ccccc3c(=O)c2Br)cc(OC)c1OC': {0: 3, 1: 0},\n",
       "             'CCCCCC1(C)OC(=O)Nc2ccc(B(O)O)cc21': {0: 14, 1: 0},\n",
       "             'CCCCCC1(C)OC(=O)Nc2ccc(-c3c(-c4cc(OC)c(OC)c(OC)c4)[nH]c4ccccc4c3=O)cc21': {0: 1,\n",
       "              1: 0},\n",
       "             'COP(=O)(OC)c1ccc2ccccc2c1Br': {0: 6, 1: 0},\n",
       "             'COc1ccc(B(O)O)c(C)c1': {0: 25, 1: 0},\n",
       "             'COc1ccc(-c2c(P(=O)(OC)OC)ccc3ccccc23)c(C)c1': {0: 1, 1: 0},\n",
       "             'O=C(F)/C=C/c1ccccc1': {0: 1, 1: 1},\n",
       "             'O=C(/C=C/c1ccccc1)c1ccccc1': {0: 7, 1: 0},\n",
       "             'N[C@@H](c1ccccc1)c1ccc(Br)cc1': {0: 1, 1: 0},\n",
       "             'N[C@@H](c1ccccc1)c1ccc(-c2ccccc2)cc1': {0: 1, 1: 0},\n",
       "             'COCn1c(-c2ccc(N3CCN(C(=O)OC(C)(C)C)CC3)cc2)c(I)c2c(Cl)ccnc21': {0: 1,\n",
       "              1: 0},\n",
       "             'CCOC(=O)c1ccc(B(O)O)cc1': {0: 61, 1: 5},\n",
       "             'CCOC(=O)c1ccc(-c2c(-c3ccc(N4CCN(C(=O)OC(C)(C)C)CC4)cc3)n(COC)c3nccc(Cl)c23)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'Cc1ccc(-c2cc(-c3cccc(Br)c3)nc(-c3ccccc3O)c2)cc1': {0: 2, 1: 0},\n",
       "             'CC(C)(C)c1ccc(O)c(B(O)O)c1': {0: 1, 1: 0},\n",
       "             'Cc1ccc(-c2cc(-c3cccc(-c4cc(C(C)(C)C)ccc4O)c3)nc(-c3ccccc3O)c2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'Nc1ncnc2c1nc(-c1ccc(Cl)cc1)n2[C@H]1C[C@H](O)[C@@H](CO)O1': {0: 10,\n",
       "              1: 0},\n",
       "             'F[B-](F)(F)c1ccccc1': {0: 275, 1: 4},\n",
       "             'Nc1ncnc2c1nc(-c1ccc(-c3ccccc3)cc1)n2[C@H]1C[C@H](O)[C@@H](CO)O1': {0: 2,\n",
       "              1: 0},\n",
       "             'COc1cc2c(c(Br)c1OC)CN(CCc1ccccc1)CC2': {0: 1, 1: 0},\n",
       "             'OB(O)c1ccccc1F': {0: 177, 1: 26},\n",
       "             'COc1cc2c(c(-c3ccccc3F)c1OC)CN(CCc1ccccc1)CC2': {0: 1, 1: 0},\n",
       "             'O=C(Cl)c1ccccc1': {0: 31, 1: 1},\n",
       "             'OB(O)/C=C/c1ccccc1': {0: 218, 1: 22},\n",
       "             'Clc1nc(-c2ccccc2)c(Cl)c(-c2ccccc2)n1': {0: 4, 1: 0},\n",
       "             'COc1ccc(-c2nc(-c3ccccc3)c(-c3ccc(OC)cc3)c(-c3ccccc3)n2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'Cn1c2c(c3ccccc31)C1(CC=CC1)c1cc(Br)ccc1-2': {0: 11, 1: 0},\n",
       "             'OB(O)c1ccc(-c2ccccc2)cc1': {0: 272, 1: 12},\n",
       "             'Cn1c2c(c3ccccc31)C1(CC=CC1)c1cc(-c3ccc(-c4ccccc4)cc3)ccc1-2': {0: 1,\n",
       "              1: 0},\n",
       "             'Brc1ccc(Br)cc1': {0: 57, 1: 5},\n",
       "             'c1ccc(-c2ccc(-c3ccccc3)cc2)cc1': {0: 22, 1: 0},\n",
       "             'Cc1ccc(S(=O)(=O)N(C(=O)c2ccc(Br)cc2)c2ccccc2)cc1': {0: 2, 1: 0},\n",
       "             'COc1ccc(-c2ccc(C(=O)N(c3ccccc3)S(=O)(=O)c3ccc(C)cc3)cc2)cc1': {0: 2,\n",
       "              1: 0},\n",
       "             'Cn1c(Br)c(C=O)c2ccccc21': {0: 5, 1: 0},\n",
       "             'Cc1ccc2ccccc2c1B(O)O': {0: 39, 1: 5},\n",
       "             'Cc1ccc2ccccc2c1-c1c(C=O)c2ccccc2n1C': {0: 1, 1: 0},\n",
       "             'O=Cc1cc(F)ccc1Br': {0: 18, 1: 0},\n",
       "             'CC(C)(C)c1cc2c(c(B3OC(C)(C)C(C)(C)O3)c1)Cc1cc(C(C)(C)C)cc(B3OC(C)(C)C(C)(C)O3)c1C2': {0: 7,\n",
       "              1: 0},\n",
       "             'CC(C)(C)c1cc2c(c(-c3ccc(F)cc3C=O)c1)Cc1cc(C(C)(C)C)cc(-c3ccc(F)cc3C=O)c1C2': {0: 1,\n",
       "              1: 0},\n",
       "             'Cc1cc(C)c(CNC(=O)c2cc(Br)nc3c2cnn3C(C)c2ccccc2)c(=O)[nH]1': {0: 0,\n",
       "              1: 1},\n",
       "             'CC1(C)C=C(B2OC(C)(C)C(C)(C)O2)CC(C)(C)N1': {0: 2, 1: 4},\n",
       "             'Cc1cc(C)c(CNC(=O)c2cc(C3=CC(C)(C)NC(C)(C)C3)nc3c2cnn3C(C)c2ccccc2)c(=O)[nH]1': {0: 0,\n",
       "              1: 1},\n",
       "             'O=C1CCc2cc(Br)ccc21': {0: 39, 1: 0},\n",
       "             'CC(C)(C)OC(=O)N1CCN(c2ccc(B3OC(C)(C)C(C)(C)O3)cc2)CC1': {0: 3,\n",
       "              1: 2},\n",
       "             'CC(C)(C)OC(=O)N1CCN(c2ccc(-c3ccc4c(c3)CCC4=O)cc2)CC1': {0: 1,\n",
       "              1: 0},\n",
       "             'O=c1cc(-c2ccccc2)oc2cc(Br)c(Br)cc12': {0: 4, 1: 0},\n",
       "             'COc1cccc(B(O)O)c1': {0: 563, 1: 34},\n",
       "             'COc1cccc(-c2cc3oc(-c4ccccc4)cc(=O)c3cc2-c2cccc(OC)c2)c1': {0: 1,\n",
       "              1: 0},\n",
       "             'Cc1ccc(S(=O)(=O)Oc2ccc(Cl)cc2)cc1': {0: 7, 1: 0},\n",
       "             'O=C(O)c1ccc(Cl)cc1': {0: 7, 1: 0},\n",
       "             'O=C(O)c1ccc(-c2ccc(F)cc2)cc1': {0: 3, 1: 0},\n",
       "             'CS(=O)(=O)c1ccncc1Nc1cc(Cl)c2nccnc2c1': {0: 4, 1: 7},\n",
       "             'CS(=O)(=O)c1ccncc1Nc1cc(-c2ccc(C(F)(F)F)cc2)c2nccnc2c1': {0: 0,\n",
       "              1: 1},\n",
       "             'COc1ccc(CSc2ccccc2Br)cc1': {0: 8, 1: 0},\n",
       "             'COc1ccc(CSc2ccccc2-c2ccc(C)cc2)cc1': {0: 1, 1: 0},\n",
       "             'C=C(OC/C=C(/C)I)C(=O)OC(C)(C)C': {0: 3, 1: 0},\n",
       "             'C=C(OCC=C(C)c1cccc2ccccc12)C(=O)OC(C)(C)C': {0: 1, 1: 0},\n",
       "             'CN(Cc1cc(Br)n(S(=O)(=O)c2ccccc2)c1)C(=O)OC(C)(C)C': {0: 1, 1: 0},\n",
       "             'CN(Cc1cc(-c2ccccc2)n(S(=O)(=O)c2ccccc2)c1)C(=O)OC(C)(C)C': {0: 1,\n",
       "              1: 0},\n",
       "             'O=C(NCc1ccco1)c1cc(S(=O)(=O)N2CCCC2)c(Br)s1': {0: 1, 1: 0},\n",
       "             'Cc1ccc(-c2sc(C(=O)NCc3ccco3)cc2S(=O)(=O)N2CCCC2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'CC(=O)c1ccc(OS(=O)(=O)C(F)(F)F)c2c1CCCC2': {0: 1, 1: 0},\n",
       "             'CC(=O)c1ccc(-c2ccc(C(C)=O)c3c2CCCC3)cc1': {0: 1, 1: 0},\n",
       "             'CCCc1ccc(-c2ccc(-c3ccc(I)cc3)cc2)c(F)c1F': {0: 3, 1: 0},\n",
       "             'CCCc1ccc(B2OCC(C)(C)CO2)cc1': {0: 55, 1: 0},\n",
       "             'CCCc1ccc(-c2ccc(-c3ccc(-c4ccc(CCC)c(F)c4F)cc3)cc2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'Clc1nc(-c2ccccc2)cc(-c2ccccc2)n1': {0: 2, 1: 0},\n",
       "             'CC1(C)OB(c2c(F)cccc2F)OC1(C)C': {0: 5, 1: 0},\n",
       "             'Fc1cccc(F)c1-c1nc(-c2ccccc2)cc(-c2ccccc2)n1': {0: 1, 1: 0},\n",
       "             'Cc1ccc(S(=O)(=O)N2C[C@H]2c2ccccc2)cc1': {0: 11, 1: 0},\n",
       "             'Cc1ccc(S(=O)(=O)NC[C@H](c2ccccc2)c2cccc(C)c2)cc1': {0: 1, 1: 0},\n",
       "             'COC(=O)c1c(OS(=O)(=O)C(F)(F)F)cc(C)c(C)c1C': {0: 4, 1: 0},\n",
       "             'COC(=O)c1c(-c2ccc(OC)cc2)cc(C)c(C)c1C': {0: 1, 1: 0},\n",
       "             'COc1cc(Br)c(OC)cc1Br': {0: 10, 1: 0},\n",
       "             'CCCCCCc1cc(B2OC(C)(C)C(C)(C)O2)c(-c2ccccc2)cc1CCCCCC': {0: 4,\n",
       "              1: 0},\n",
       "             'CCCCCCc1cc(-c2ccccc2)c(-c2cc(OC)c(-c3cc(CCCCCC)c(CCCCCC)cc3-c3ccccc3)cc2OC)cc1CCCCCC': {0: 1,\n",
       "              1: 0},\n",
       "             'CN(C)S(=O)(=O)c1cccc(-c2ccnc3c(I)c(-c4ccncc4)nn23)c1': {0: 0,\n",
       "              1: 2},\n",
       "             'OB(O)c1cccc2[nH]ccc12': {0: 20, 1: 8},\n",
       "             'CN(C)S(=O)(=O)c1cccc(-c2ccnc3c(-c4cccc5[nH]ccc45)c(-c4ccncc4)nn23)c1': {0: 0,\n",
       "              1: 1},\n",
       "             'COc1c(NCC2CCCCC2)cnc2ccc(Br)cc12': {0: 0, 1: 1},\n",
       "             'COc1ncc(B2OC(C)(C)C(C)(C)O2)cc1NS(=O)(=O)c1ccc(F)cc1F': {0: 21,\n",
       "              1: 41},\n",
       "             'COc1ncc(-c2ccc3ncc(NCC4CCCCC4)c(OC)c3c2)cc1NS(=O)(=O)c1ccc(F)cc1F': {0: 0,\n",
       "              1: 1},\n",
       "             'Cn1c(=O)sc2ccc(Br)cc21': {0: 5, 1: 0},\n",
       "             'N#Cc1cccc(B(O)O)c1': {0: 118, 1: 23},\n",
       "             'Cn1c(=O)sc2ccc(-c3cccc(C#N)c3)cc21': {0: 1, 1: 0},\n",
       "             'Cc1c(C)c(Br)c(C)c(C)c1O': {0: 1, 1: 0},\n",
       "             'Cc1c(C)c(-c2cccc(C=O)c2)c(C)c(C)c1O': {0: 1, 1: 0},\n",
       "             'Brc1c2ccccc2c(-c2c3ccccc3c(Br)c3ccccc23)c2ccccc12': {0: 9, 1: 0},\n",
       "             'Fc1ccccc1-c1c2ccccc2c(-c2c3ccccc3c(-c3ccccc3F)c3ccccc23)c2ccccc12': {0: 1,\n",
       "              1: 0},\n",
       "             'Cn1nc(I)c2cccnc21': {0: 7, 1: 0},\n",
       "             'COc1cccc(-c2nn(C)c3ncccc23)c1': {0: 1, 1: 0},\n",
       "             'COc1cc(Br)c2c(c1)N(C)C(=O)C2': {0: 3, 1: 0},\n",
       "             'Cc1ccc(C=O)c(B(O)O)c1': {0: 8, 1: 0},\n",
       "             'COc1cc2c3c(cc4ccc(C)cc4c3c1)C(=O)N2C': {0: 1, 1: 0},\n",
       "             'COc1ccc(C#Cc2ccc(Br)cc2)cc1': {0: 5, 1: 0},\n",
       "             'OB(O)c1ccc(O)cc1': {0: 254, 1: 46},\n",
       "             'COc1ccc(C#Cc2ccc(-c3ccc(O)cc3)cc2)cc1': {0: 1, 1: 0},\n",
       "             'O=C(c1ccc(Br)s1)C(F)(F)F': {0: 1, 1: 9},\n",
       "             'OB(O)c1ccccn1': {0: 12, 1: 7},\n",
       "             'O=C(c1ccc(-c2ccccn2)s1)C(F)(F)F': {0: 0, 1: 1},\n",
       "             'Cc1ccc(O)c(Br)c1': {0: 7, 1: 1},\n",
       "             'COc1cc(OC)cc(B(O)O)c1': {0: 106, 1: 2},\n",
       "             'COc1cc(OC)cc(-c2cc(C)ccc2O)c1': {0: 1, 1: 0},\n",
       "             'Nc1ncc(Br)c(-c2c[nH]c3ccccc23)n1': {0: 0, 1: 4},\n",
       "             'Nc1ncc(-c2ccc(C(F)(F)F)cc2)c(-c2c[nH]c3ccccc23)n1': {0: 0, 1: 1},\n",
       "             'COC(=O)C1=C(OS(=O)(=O)C(F)(F)F)CCCCC1': {0: 9, 1: 0},\n",
       "             'COC(=O)C1=C(c2ccc3ccccc3c2)CCCCC1': {0: 1, 1: 0},\n",
       "             'Ic1ccc(I)cc1': {0: 14, 1: 4},\n",
       "             'Cc1sc(B(O)O)c(C)c1Br': {0: 1, 1: 1},\n",
       "             'Cc1sc(-c2ccc(-c3sc(C)c(Br)c3C)cc2)c(C)c1Br': {0: 0, 1: 1},\n",
       "             'Clc1nccc2c1sc1ccccc12': {0: 1, 1: 0},\n",
       "             'c1ccc(-c2nccc3c2sc2ccccc23)cc1': {0: 1, 1: 0},\n",
       "             'CCCCCCc1cc(Br)sc1-c1nc2sc(-c3sc(Br)cc3CCCCCC)nc2s1': {0: 1,\n",
       "              1: 0},\n",
       "             'CCCCCCc1cc(-c2ccccc2)sc1-c1nc2sc(-c3sc(-c4ccccc4)cc3CCCCCC)nc2s1': {0: 1,\n",
       "              1: 0},\n",
       "             'COc1ccc(C(=Cc2ccccc2)OC(C)=O)cc1': {0: 1, 1: 0},\n",
       "             'c1ccc(B2OB(c3ccccc3)OB(c3ccccc3)O2)cc1': {0: 47, 1: 2},\n",
       "             'COc1ccc(/C(=C\\\\c2ccccc2)c2ccccc2)cc1': {0: 1, 1: 0},\n",
       "             'COc1ccc(I)cc1': {0: 131, 1: 8},\n",
       "             'COc1ccc(-c2ccc(C(F)(F)F)cc2)cc1': {0: 13, 1: 0},\n",
       "             'CC(C)(C)OC(=O)N[C@@H](Cc1ccc(Br)cc1)C(=O)O': {0: 8, 1: 0},\n",
       "             'CC(C)(C)OC(=O)N[C@@H](Cc1ccc(-c2ccc(C(F)(F)F)cc2)cc1)C(=O)O': {0: 1,\n",
       "              1: 0},\n",
       "             'CCCCCCC1=C[B-]2(OC(C)(C)C(C)(C)O2)O[Si]1(C)C': {0: 1, 1: 0},\n",
       "             'CCCCCCC(=Cc1ccc(OC)cc1)[Si](C)(C)OB1OC(C)(C)C(C)(C)O1': {0: 1,\n",
       "              1: 0},\n",
       "             'O=C(O)c1cccc(Br)c1': {0: 30, 1: 1},\n",
       "             'OB(O)c1cccc(F)c1': {0: 204, 1: 26},\n",
       "             'O=C(O)c1cccc(-c2cccc(F)c2)c1': {0: 1, 1: 0},\n",
       "             'COC(=O)c1cc(Br)c(C(=O)OC)cc1Br': {0: 7, 1: 0},\n",
       "             'CCCCCCCCc1ccc(C2(c3ccc(CCCCCCCC)cc3)c3ccccc3-c3ccc(B4OC(C)(C)C(C)(C)O4)cc32)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'CCCCCCCCc1ccc(C2(c3ccc(CCCCCCCC)cc3)c3ccccc3-c3ccc(-c4cc(C(=O)OC)c(-c5ccc6c(c5)C(c5ccc(CCCCCCCC)cc5)(c5ccc(CCCCCCCC)cc5)c5ccccc5-6)cc4C(=O)OC)cc32)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'COc1ccc2ccc(Br)cc2c1-c1c(OC)ccc2ccc(Br)cc12': {0: 1, 1: 0},\n",
       "             'CC1(C)OB(c2cccc(P(=O)(c3ccccc3)c3ccccc3)c2)OC1(C)C': {0: 1,\n",
       "              1: 0},\n",
       "             'COc1ccc2ccc(-c3cccc(P(=O)(c4ccccc4)c4ccccc4)c3)cc2c1-c1c(OC)ccc2ccc(-c3cccc(P(=O)(c4ccccc4)c4ccccc4)c3)cc12': {0: 1,\n",
       "              1: 0},\n",
       "             'Brc1nccc2ccncc12': {0: 2, 1: 0},\n",
       "             'COC(=O)c1ccccc1B(O)O': {0: 27, 1: 7},\n",
       "             'COC(=O)c1ccccc1-c1nccc2ccncc12': {0: 1, 1: 0},\n",
       "             'Nc1nc2c(Br)cccn2n1': {0: 6, 1: 0},\n",
       "             'Cc1cc(F)ccc1B(O)O': {0: 19, 1: 1},\n",
       "             'Cc1cc(F)ccc1-c1cccn2nc(N)nc12': {0: 1, 1: 0},\n",
       "             'O=S(=O)(Nc1ccccc1)c1ccc(Br)cc1': {0: 1, 1: 0},\n",
       "             'C[C@@H]1CCCN1CCc1ccc(B(O)O)cc1': {0: 8, 1: 3},\n",
       "             'C[C@@H]1CCCN1CCc1ccc(-c2ccc(S(=O)(=O)Nc3ccccc3)cc2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'CCCCNC(=O)c1ccc(Br)cc1': {0: 1, 1: 0},\n",
       "             'CCCCNC(=O)c1ccc(-c2cc(OC)c(OC)c(OC)c2)cc1': {0: 1, 1: 0},\n",
       "             'CCOP(=O)(OCC)OC1=C(Br)CC1': {0: 11, 1: 0},\n",
       "             'CCOP(=O)(OCC)OC1=C(c2ccc(F)cc2)CC1': {0: 1, 1: 0},\n",
       "             'C=C(Br)CCO': {0: 11, 1: 0},\n",
       "             'F[B-](F)(F)/C=C/c1ccccc1': {0: 62, 1: 1},\n",
       "             'C=C(/C=C/c1ccccc1)CCO': {0: 1, 1: 0},\n",
       "             'BrC(Br)=c1c2ccccc2c(=C(Br)Br)c2nc(-c3ccccc3)c(-c3ccccc3)nc12': {0: 1,\n",
       "              1: 0},\n",
       "             'COc1ccc(C(c2ccc(OC)cc2)=c2c3ccccc3c(=C(c3ccc(OC)cc3)c3ccc(OC)cc3)c3nc(-c4ccccc4)c(-c4ccccc4)nc23)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'CC(O[C@@H]1COc2nc([N+](=O)[O-])cn2C1)c1ccc(Br)cc1': {0: 4, 1: 0},\n",
       "             'CC(O[C@@H]1COc2nc([N+](=O)[O-])cn2C1)c1ccc(-c2ccc(F)cc2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'COc1cc(OC)cc(OS(=O)(=O)c2c(F)c(F)c(F)c(F)c2F)c1': {0: 22, 1: 0},\n",
       "             'CC(C)(O)c1ccc(B(O)O)cc1': {0: 3, 1: 1},\n",
       "             'COc1cc(OC)cc(-c2ccc(C(C)(C)O)cc2)c1': {0: 1, 1: 0},\n",
       "             'Cc1ccc(S(=O)(=O)Oc2ccc3ccc(OS(=O)(=O)c4ccc(C)cc4)cc3c2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'c1ccc(-c2ccc3ccc(-c4ccccc4)cc3c2)cc1': {0: 1, 1: 0},\n",
       "             'N#Cc1ccc(CBr)cc1': {0: 7, 1: 0},\n",
       "             'COc1ccc([B-](F)(F)F)cc1': {0: 72, 1: 5},\n",
       "             'COc1ccc(Cc2ccc(C#N)cc2)cc1': {0: 4, 1: 0},\n",
       "             'C=CCC#CC(OC(=O)OC)c1ccc(Cl)cc1': {0: 1, 1: 0},\n",
       "             'C=CCC(=C=Cc1ccc(Cl)cc1)c1ccccc1': {0: 2, 1: 0},\n",
       "             'Brc1csc(-c2cn([C@H]3CN4CCC3CC4)nn2)c1': {0: 2, 1: 0},\n",
       "             'OCc1ccc(B(O)O)cc1': {0: 124, 1: 14},\n",
       "             'OCc1ccc(-c2csc(-c3cn([C@H]4CN5CCC4CC5)nn3)c2)cc1': {0: 1, 1: 0},\n",
       "             'Brc1ccc(-c2cc(-c3ccccn3)nc(-c3ccccn3)c2)cc1': {0: 8, 1: 0},\n",
       "             'CC1(C)OB(c2cccc(B3OC(C)(C)C(C)(C)O3)c2)OC1(C)C': {0: 5, 1: 5},\n",
       "             'c1ccc(-c2cc(-c3ccc(-c4cccc(-c5ccc(-c6cc(-c7ccccn7)nc(-c7ccccn7)c6)cc5)c4)cc3)cc(-c3ccccn3)n2)nc1': {0: 1,\n",
       "              1: 0},\n",
       "             'COc1ccc2c(OS(=O)(=O)C(F)(F)F)cccc2c1': {0: 1, 1: 0},\n",
       "             'O=Cc1sccc1B(O)O': {0: 22, 1: 0},\n",
       "             'COc1ccc2c(-c3ccsc3C=O)cccc2c1': {0: 1, 1: 0},\n",
       "             'O=C1C(Br)=C(Br)C(=O)c2ccccc21': {0: 3, 1: 1},\n",
       "             'O=[N+]([O-])c1ccccc1B(O)O': {0: 56, 1: 5},\n",
       "             'O=C1C(c2ccccc2[N+](=O)[O-])=C(c2ccccc2[N+](=O)[O-])C(=O)c2ccccc21': {0: 0,\n",
       "              1: 1},\n",
       "             'O=C(c1ccc(Br)cc1)c1cc2cc(Br)ccc2o1': {0: 5, 1: 0},\n",
       "             'O=C(c1ccc(-c2ccccc2)cc1)c1cc2cc(-c3ccccc3)ccc2o1': {0: 1, 1: 0},\n",
       "             'Brc1sc(Br)c(Br)c1Br': {0: 26, 1: 1},\n",
       "             'OB(O)c1cccs1': {0: 576, 1: 65},\n",
       "             'c1csc(-c2sc(-c3cccs3)c(-c3cccs3)c2-c2cccs2)c1': {0: 1, 1: 0},\n",
       "             'CCOC(=O)C1CCCN(CCC#Cc2ccccc2Br)C1': {0: 11, 1: 0},\n",
       "             'CCOC(=O)C1CCCN(CCC#Cc2ccccc2-c2ccccc2)C1': {0: 1, 1: 0},\n",
       "             'Cn1cc(I)c(=O)c2ccccc21': {0: 7, 1: 2},\n",
       "             'Cn1cc(-c2cccs2)c(=O)c2ccccc21': {0: 1, 1: 0},\n",
       "             'Cc1cc(Oc2ccccc2)c(Cl)c(=O)o1': {0: 1, 1: 0},\n",
       "             'Cc1cc(Oc2ccccc2)c(-c2ccccc2)c(=O)o1': {0: 1, 1: 0},\n",
       "             'COc1ncc(I)c(OC)n1': {0: 4, 1: 1},\n",
       "             'OB(O)c1cc(F)cnc1Cl': {0: 0, 1: 1},\n",
       "             'COc1ncc(-c2cc(F)cnc2Cl)c(OC)n1': {0: 0, 1: 1},\n",
       "             'CC1(C)CCN(CCOc2ccc3[nH]ccc3c2)c2ccc(Br)cc21': {0: 2, 1: 0},\n",
       "             'OB(O)c1ccco1': {0: 310, 1: 42},\n",
       "             'CC1(C)CCN(CCOc2ccc3[nH]ccc3c2)c2ccc(-c3ccco3)cc21': {0: 1, 1: 0},\n",
       "             'Nc1nc(N2CCOCC2)c2nc(Cl)ccc2n1': {0: 0, 1: 1},\n",
       "             'O=C(NC1CC1)c1ccc(B(O)O)cc1': {0: 5, 1: 1},\n",
       "             'Nc1nc(N2CCOCC2)c2nc(-c3ccc(C(=O)NC4CC4)cc3)ccc2n1': {0: 0, 1: 1},\n",
       "             'Clc1ccccc1-n1ccc2cccnc21': {0: 2, 1: 0},\n",
       "             'c1ccc(-c2ccccc2-n2ccc3cccnc32)cc1': {0: 1, 1: 0},\n",
       "             'COC(=O)[C@]12CS(=O)(=O)[C@H](C(C)(C)C)N1C(=O)C=C2OS(C)(=O)=O': {0: 0,\n",
       "              1: 3},\n",
       "             'COC(=O)[C@]12CS(=O)(=O)[C@H](C(C)(C)C)N1C(=O)C=C2c1ccco1': {0: 0,\n",
       "              1: 1},\n",
       "             'Fc1cc(Br)ccc1Cl': {0: 1, 1: 0},\n",
       "             'Clc1ncc(Cl)c(NCCCB2O[C@H]3CCC[C@@H]2CCC3)n1': {0: 3, 1: 0},\n",
       "             'Fc1cc(CCCNc2nc(Cl)ncc2Cl)ccc1Cl': {0: 1, 1: 0},\n",
       "             'Clc1ccc(Cl)cc1': {0: 11, 1: 0},\n",
       "             'Cc1cccc(C)c1B(O)O': {0: 157, 1: 16},\n",
       "             'Cc1cccc(C)c1-c1ccc(-c2c(C)cccc2C)cc1': {0: 1, 1: 1},\n",
       "             'Oc1ccc2cc(Br)ccc2c1': {0: 16, 1: 0},\n",
       "             'Oc1ccc(-c2ccc3cc(O)ccc3c2)cc1': {0: 1, 1: 0},\n",
       "             'Cn1c(Br)c(Br)c2ccccc21': {0: 21, 1: 1},\n",
       "             'COc1ccc(-c2c(Br)c3ccccc3n2C)cc1OC': {0: 1, 1: 0},\n",
       "             'O=Cc1cccc(Br)c1': {0: 39, 1: 1},\n",
       "             'O=Cc1cccc(-c2ccc(F)cc2)c1': {0: 1, 1: 0},\n",
       "             'O=C(Nc1ccc(Br)c2cccnc12)c1ccccc1': {0: 1, 1: 0},\n",
       "             'O=C(Nc1ccc(-c2ccccc2)c2cccnc12)c1ccccc1': {0: 1, 1: 0},\n",
       "             'Cn1nnn(-c2cc([N+](=O)[O-])c(F)cc2Br)c1=O': {0: 1, 1: 0},\n",
       "             'F[B-](F)(F)C1CC1': {0: 36, 1: 5},\n",
       "             'Cn1nnn(-c2cc([N+](=O)[O-])c(F)cc2C2CC2)c1=O': {0: 1, 1: 0},\n",
       "             'O=Cc1cc2ccccc2nc1Cl': {0: 12, 1: 0},\n",
       "             'Cc1cccc(-c2nc3ccccc3cc2C=O)c1': {0: 1, 1: 0},\n",
       "             'COc1cc(OC(=O)C(C)(C)C)cc(OS(=O)(=O)c2ccc(C)cc2)c1': {0: 1, 1: 0},\n",
       "             'COc1cc(OC(=O)C(C)(C)C)cc(-c2ccccc2)c1': {0: 1, 1: 0},\n",
       "             'CC(C)(C)c1cc(Br)c(I)c(Br)c1': {0: 1, 1: 1},\n",
       "             'OB(O)c1ccc(B(O)O)cc1': {0: 58, 1: 12},\n",
       "             'CC(C)(C)c1cc(Br)c(-c2ccc(-c3c(Br)cc(C(C)(C)C)cc3Br)cc2)c(Br)c1': {0: 0,\n",
       "              1: 1},\n",
       "             'C=C1CC(=O)N[C@@H](C)C(=O)N(C)[C@@H](C(C)C)C(=O)NCC(=O)O[C@@H](C(C)=CI)[C@@H]1C': {0: 2,\n",
       "              1: 0},\n",
       "             'CC(C)(C)C=CB1OC(C)(C)C(C)(C)O1': {0: 1, 1: 0},\n",
       "             'C=C1CC(=O)N[C@@H](C)C(=O)N(C)[C@@H](C(C)C)C(=O)NCC(=O)O[C@@H](/C(C)=C/C=C/C(C)(C)C)[C@@H]1C': {0: 1,\n",
       "              1: 0},\n",
       "             'Nc1nc(Cl)cc(N2CCCC2C2CC2)n1': {0: 0, 1: 1},\n",
       "             'CC(=O)n1nc2c3c(cc(B4OC(C)(C)C(C)(C)O4)cc31)Oc1ccccc1N2': {0: 0,\n",
       "              1: 13},\n",
       "             'Nc1nc(-c2cc3c4c(c2)Oc2ccccc2N=C4NN3)cc(N2CCCC2C2CC2)n1': {0: 0,\n",
       "              1: 1},\n",
       "             'CN1CCCN(C2CCC(Oc3cc(=O)n(C)cc3Br)CC2)C1=O': {0: 1, 1: 0},\n",
       "             'Cc1cc(F)cc(C)c1Oc1ccc(C(C)(C)O)cc1B1OC(C)(C)C(C)(C)O1': {0: 16,\n",
       "              1: 4},\n",
       "             'Cc1cc(F)cc(C)c1Oc1ccc(C(C)(C)O)cc1-c1cn(C)c(=O)cc1OC1CCC(N2CCCN(C)C2=O)CC1': {0: 1,\n",
       "              1: 0},\n",
       "             'Brc1cnc2nnc(Cc3ccc4ncccc4c3)n2c1': {0: 0, 1: 1},\n",
       "             'c1ccc(-c2cnc3nnc(Cc4ccc5ncccc5c4)n3c2)cc1': {0: 0, 1: 1},\n",
       "             'Brc1ccc(Br)nc1': {0: 32, 1: 4},\n",
       "             'OB(O)c1ccc(C(=C(c2ccccc2)c2ccccc2)c2ccccc2)cc1': {0: 21, 1: 0},\n",
       "             'c1ccc(C(=C(c2ccccc2)c2ccc(-c3ccc(-c4ccc(C(=C(c5ccccc5)c5ccccc5)c5ccccc5)cc4)nc3)cc2)c2ccccc2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'CCOC(=O)c1ccc(I)cc1': {0: 16, 1: 1},\n",
       "             'CC1(C)OB(/C(=C(\\\\B2OC(C)(C)C(C)(C)O2)[Si](C)(C)C)c2ccccc2)OC1(C)C': {0: 8,\n",
       "              1: 0},\n",
       "             'CCOC(=O)c1ccc(/C(=C(\\\\B2OC(C)(C)C(C)(C)O2)c2ccccc2)[Si](C)(C)C)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'Clc1nc(Cl)nc(-c2ccccc2)n1': {0: 25, 1: 1},\n",
       "             'OB(O)c1cccc2c1sc1ccccc12': {0: 54, 1: 6},\n",
       "             'Clc1nc(-c2ccccc2)nc(-c2cccc3c2sc2ccccc23)n1': {0: 1, 1: 0},\n",
       "             'O=C1Nc2cccc(I)c2C1=O': {0: 25, 1: 1},\n",
       "             'OB(O)c1cc(F)cc(F)c1': {0: 83, 1: 14},\n",
       "             'O=C1Nc2cccc(-c3cc(F)cc(F)c3)c2C1=O': {0: 1, 1: 0},\n",
       "             'COc1cc(C=C(F)Br)cc(OC)c1': {0: 2, 1: 0},\n",
       "             'COc1ccc(/C(F)=C/c2cc(OC)cc(OC)c2)cc1': {0: 2, 1: 0},\n",
       "             'Cc1nc(C)c([C@H](OC(C)(C)C)C(=O)OC(C)C)c(N2CCC3(CCC3)CC2)c1Br': {0: 3,\n",
       "              1: 0},\n",
       "             'OB(O)c1ccc(OCCc2ccc(F)cc2)cc1': {0: 11, 1: 2},\n",
       "             'Cc1nc(C)c([C@H](OC(C)(C)C)C(=O)OC(C)C)c(N2CCC3(CCC3)CC2)c1-c1ccc(OCCc2ccc(F)cc2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'CCc1cc2c(cc1O[Si](C)(C)C(C)(C)C)CC[C@@H]1[C@@H]2CC[C@]2(C)C(=O)CC[C@@H]12': {0: 3,\n",
       "              1: 0},\n",
       "             'OB(O)c1ccnc(Cl)c1': {0: 15, 1: 4},\n",
       "             'CCc1cc2c(cc1O[Si](C)(C)C(C)(C)C)CC[C@@H]1[C@@H]2CC[C@]2(C)C(c3ccnc(Cl)c3)=CC[C@@H]12': {0: 1,\n",
       "              1: 0},\n",
       "             'CC(C)(C)c1cc(I)cc(I)c1': {0: 1, 1: 0},\n",
       "             'CC1(C)OB(c2ccc(-c3cccs3)s2)OC1(C)C': {0: 31, 1: 2},\n",
       "             'CC(C)(C)c1cc(-c2ccc(-c3cccs3)s2)cc(-c2ccc(-c3cccs3)s2)c1': {0: 1,\n",
       "              1: 0},\n",
       "             'COCOC(=O)/C=C(\\\\I)CCO': {0: 7, 1: 0},\n",
       "             'CC1(C)OB(/C=C/CCCCCCc2cccc(CCCCCCCCO[Si](C)(C)C(C)(C)C)c2)OC1(C)C': {0: 1,\n",
       "              1: 0},\n",
       "             'COCOC(=O)/C=C(\\\\C=C\\\\CCCCCCc1cccc(CCCCCCCCO[Si](C)(C)C(C)(C)C)c1)CCO': {0: 1,\n",
       "              1: 0},\n",
       "             'CCOC(=O)c1cc(Br)c(C#N)[nH]1': {0: 9, 1: 0},\n",
       "             'CCOC(=O)c1cc(-c2ccc(Cl)cc2)c(C#N)[nH]1': {0: 1, 1: 0},\n",
       "             'Brc1nccs1': {0: 30, 1: 1},\n",
       "             'OB(O)c1ccc(-c2ccc(N(c3ccccc3)c3ccccc3)cc2)cc1': {0: 6, 1: 1},\n",
       "             'c1ccc(N(c2ccccc2)c2ccc(-c3ccc(-c4nccs4)cc3)cc2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'O=C1N(Cc2ccccc2)[C@H](c2ccccc2)[C@]1(F)Br': {0: 7, 1: 0},\n",
       "             'Cc1cccc(B2C3CCCC2CCC3)c1': {0: 1, 1: 0},\n",
       "             'Cc1cccc([C@]2(F)C(=O)N(Cc3ccccc3)[C@H]2c2ccccc2)c1': {0: 1,\n",
       "              1: 0},\n",
       "             'Brc1ccc2nc(OCc3cccc4ccccc34)c(I)cc2c1': {0: 17, 1: 0},\n",
       "             'CC1(C)OB(c2ccc(N3CCCC3)cc2)OC1(C)C': {0: 2, 1: 0},\n",
       "             'Brc1ccc2nc(OCc3cccc4ccccc34)c(-c3ccc(N4CCCC4)cc3)cc2c1': {0: 1,\n",
       "              1: 0},\n",
       "             'FC(F)(F)c1cc(-c2cc([S+]3CCCC3)c3cccccc2-3)cc(C(F)(F)F)c1': {0: 0,\n",
       "              1: 1},\n",
       "             'O=Cc1ccc(B(O)O)o1': {0: 82, 1: 14},\n",
       "             'O=Cc1ccc(-c2cc(-c3cc(C(F)(F)F)cc(C(F)(F)F)c3)c3cccccc2-3)o1': {0: 0,\n",
       "              1: 1},\n",
       "             'O=C(c1ccccc1Br)c1ccccc1Br': {0: 1, 1: 0},\n",
       "             'O=C(c1ccccc1-c1ccccc1)c1ccccc1-c1ccccc1': {0: 1, 1: 0},\n",
       "             'CC(C)[Si](O[C@@H]1c2ccccc2C(OS(=O)(=O)C(F)(F)F)=C2COC[C@H]21)(C(C)C)C(C)C': {0: 1,\n",
       "              1: 0},\n",
       "             'CC(C)[Si](O[C@@H]1c2ccccc2C(c2ccccc2)=C2COC[C@H]21)(C(C)C)C(C)C': {0: 1,\n",
       "              1: 0},\n",
       "             'CC(C)(C)C(=O)OCn1cnc2c(Cl)nc(I)nc21': {0: 1, 1: 0},\n",
       "             'CC(C)(C)C(=O)OCn1cnc2c(Cl)nc(-c3ccccc3)nc21': {0: 1, 1: 0},\n",
       "             'Nc1ncn(Cc2ccccc2)c2nc(Br)nc1-2': {0: 4, 1: 0},\n",
       "             'O=[N+]([O-])c1cccc(B(O)O)c1': {0: 294, 1: 35},\n",
       "             'Nc1ncn(Cc2ccccc2)c2nc(-c3cccc([N+](=O)[O-])c3)nc1-2': {0: 1,\n",
       "              1: 0},\n",
       "             'Nc1ccccc1C(=O)c1ccc(Br)cc1': {0: 2, 1: 0},\n",
       "             'COc1ccc(-c2ccc(C(=O)c3ccccc3N)cc2)cc1': {0: 1, 1: 0},\n",
       "             'CC(/C=C(\\\\I)CO)CO[Si](c1ccccc1)(c1ccccc1)C(C)(C)C': {0: 1, 1: 0},\n",
       "             'OB(O)c1ccc2[nH]ccc2c1': {0: 95, 1: 11},\n",
       "             'CC(/C=C(/CO)c1ccc2[nH]ccc2c1)CO[Si](c1ccccc1)(c1ccccc1)C(C)(C)C': {0: 1,\n",
       "              1: 0},\n",
       "             'COC(=O)c1ccc(Cl)cc1': {0: 23, 1: 2},\n",
       "             'COC(=O)c1ccc(-c2c(C)cccc2C)cc1': {0: 3, 1: 0},\n",
       "             'C[C@H](NC(=O)OCc1ccccc1)c1nc2ccc(F)cn2c1I': {0: 4, 1: 1},\n",
       "             'CS(=O)(=O)c1cccc(B(O)O)c1': {0: 23, 1: 9},\n",
       "             'C[C@H](NC(=O)OCc1ccccc1)c1nc2ccc(F)cn2c1-c1cccc(S(C)(=O)=O)c1': {0: 1,\n",
       "              1: 0},\n",
       "             'NNc1ccccc1': {0: 35, 1: 0},\n",
       "             'Fc1ccccc1-c1ccccc1': {0: 14, 1: 0},\n",
       "             'CC(C)Nc1ncc(Br)n(CC(=O)OC(C)(C)C)c1=O': {0: 8, 1: 1},\n",
       "             'COc1ccc(-c2cnc(NC(C)C)c(=O)n2CC(=O)OC(C)(C)C)cc1': {0: 1, 1: 0},\n",
       "             'Clc1ccc(N2CCOCC2)nn1': {0: 1, 1: 1},\n",
       "             'c1cc2cc(-c3ccc(N4CCOCC4)nn3)ccc2[nH]1': {0: 1, 1: 0},\n",
       "             'NC(=O)Nc1sc(Br)cc1C(N)=O': {0: 3, 1: 5},\n",
       "             'CCC(CC)(NCc1cccc(B2OC(C)(C)C(C)(C)O2)c1)C(=O)OC1CCCC1': {0: 0,\n",
       "              1: 1},\n",
       "             'CCC(CC)(NCc1cccc(-c2cc(C(N)=O)c(NC(N)=O)s2)c1)C(=O)OC1CCCC1': {0: 0,\n",
       "              1: 1},\n",
       "             'COC(=O)C1=C(OS(=O)(=O)C(F)(F)F)CCC1': {0: 16, 1: 0},\n",
       "             'COc1ccc([B-](O)(O)O)cc1': {0: 10, 1: 0},\n",
       "             'COC(=O)C1=C(c2ccc(OC)cc2)CCC1': {0: 2, 1: 0},\n",
       "             'COc1cccc(Cl)c1': {0: 16, 1: 1},\n",
       "             'COc1ccc(-c2cccc(OC)c2)cc1': {0: 10, 1: 0},\n",
       "             'O=Cc1ccc(Br)s1': {0: 58, 1: 3},\n",
       "             'CCCCC(CC)CC1(CC(CC)CCCC)C2=C(c3ccc(B4OC(C)(C)C(C)(C)O4)cc31)C(CC(CC)CCCC)(CC(CC)CCCC)c1cc(B3OC(C)(C)C(C)(C)O3)ccc12': {0: 1,\n",
       "              1: 0},\n",
       "             'CCCCC(CC)CC1(CC(CC)CCCC)C2=C(c3ccc(-c4ccc(C=O)s4)cc31)C(CC(CC)CCCC)(CC(CC)CCCC)c1cc(-c3ccc(C=O)s3)ccc12': {0: 1,\n",
       "              1: 0},\n",
       "             'O=CC1=C(Cl)c2c(ccc3ccccc23)CC1': {0: 3, 1: 0},\n",
       "             'Cc1ccccc1B(O)O': {0: 686, 1: 23},\n",
       "             'Cc1ccccc1C1=C(C=O)CCc2ccc3ccccc3c21': {0: 1, 1: 0},\n",
       "             'O=C(O)c1ccc(Br)cc1': {0: 33, 1: 0},\n",
       "             'OB(O)c1ccc(N(c2ccccc2)c2ccccc2)cc1': {0: 276, 1: 4},\n",
       "             'O=C(O)c1ccc(-c2ccc(N(c3ccccc3)c3ccccc3)cc2)cc1': {0: 1, 1: 0},\n",
       "             'CC(C)Oc1ccccc1Cl': {0: 1, 1: 0},\n",
       "             'CC(C)Oc1ccccc1-c1ccccc1': {0: 1, 1: 0},\n",
       "             'CCCCCCCCCCCC(=O)Nc1ccc(Br)cc1': {0: 4, 1: 0},\n",
       "             'CCCCCCCCCCCC(=O)Nc1ccc(B2OC(C)(C)C(C)(C)O2)cc1': {0: 1, 1: 0},\n",
       "             'CCCCCCCCCCCC(=O)Nc1ccc(-c2ccc(NC(=O)CCCCCCCCCCC)cc2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'O=Cc1cc(-c2cccs2)ccc1OS(=O)(=O)C(F)(F)F': {0: 1, 1: 0},\n",
       "             'O=Cc1ccccc1B(O)O': {0: 308, 1: 10},\n",
       "             'O=Cc1ccccc1-c1ccc(-c2ccsc2)cc1C=O': {0: 1, 1: 0},\n",
       "             'Brc1nc(-c2ccco2)n(Cc2ccccc2)c1Br': {0: 1, 1: 0},\n",
       "             'COc1ccc(-c2nc(-c3ccco3)n(Cc3ccccc3)c2-c2ccc(OC)cc2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'CCn1c(=O)c(Br)cn(-c2ccc(Cl)cc2)c1=O': {0: 0, 1: 6},\n",
       "             'CCn1c(=O)c(-c2ccc(OC)c(OC)c2)cn(-c2ccc(Cl)cc2)c1=O': {0: 0,\n",
       "              1: 1},\n",
       "             'BrC#Cc1ccccc1': {0: 23, 1: 0},\n",
       "             'CSc1ccc(B(O)O)cc1': {0: 117, 1: 12},\n",
       "             'CSc1ccc(C#Cc2ccccc2)cc1': {0: 1, 1: 0},\n",
       "             'CO/N=C(\\\\CCl)c1ccccc1': {0: 9, 1: 0},\n",
       "             'CO/N=C(\\\\Cc1ccc(OC)cc1)c1ccccc1': {0: 1, 1: 0},\n",
       "             'O=c1ccn(-c2cccc(C(F)(F)F)c2)nc1Br': {0: 0, 1: 2},\n",
       "             'OB(O)c1ccoc1-c1ccccc1': {0: 0, 1: 1},\n",
       "             'O=c1ccn(-c2cccc(C(F)(F)F)c2)nc1-c1ccoc1-c1ccccc1': {0: 0, 1: 1},\n",
       "             'C/C(=C\\\\[C@@H](OC(=O)C(C)(C)C)c1ccccc1)CCC1OC1(C)C': {0: 1,\n",
       "              1: 0},\n",
       "             'COc1cccc(B2OB(c3cccc(OC)c3)OB(c3cccc(OC)c3)O2)c1': {0: 12, 1: 0},\n",
       "             'COc1cccc(C(C)(/C=C/c2ccccc2)CCC2OC2(C)C)c1': {0: 1, 1: 0},\n",
       "             'COc1ccc(-c2oc3cc(OC)cc(C=O)c3c2I)cc1': {0: 4, 1: 0},\n",
       "             'COc1ccc(-c2oc3cc(OC)cc(C=O)c3c2-c2cc(OC)c(OC)c(OC)c2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'COCn1ncc(Cl)c(OC)c1=O': {0: 1, 1: 0},\n",
       "             'COCn1ncc(-c2ccc(F)cc2)c(OC)c1=O': {0: 1, 1: 0},\n",
       "             'Cc1ccc(C)c(Br)c1': {0: 10, 1: 0},\n",
       "             'Cc1ccc(C)c(-c2ccc(Cl)cc2)c1': {0: 1, 1: 0},\n",
       "             'Brc1cccc(-c2cccc(-c3cccc(Br)n3)n2)n1': {0: 2, 1: 0},\n",
       "             'CSc1ccccc1B(O)O': {0: 38, 1: 4},\n",
       "             'CSc1ccccc1-c1cccc(-c2cccc(-c3cccc(-c4ccccc4SC)n3)n2)n1': {0: 1,\n",
       "              1: 0},\n",
       "             'Clc1cc(NCc2ccccc2)ncn1': {0: 8, 1: 0},\n",
       "             'c1ccc(CNc2cc(-c3ccsc3)ncn2)cc1': {0: 1, 1: 0},\n",
       "             'COc1cc2c(Cl)cc(Cl)nc2cc1OCCCN1CCOCC1': {0: 1, 1: 0},\n",
       "             'CCc1ccc(B2OC(C)(C)C(C)(C)O2)o1': {0: 4, 1: 0},\n",
       "             'CCc1ccc(-c2cc(Cl)c3cc(OC)c(OCCCN4CCOCC4)cc3n2)o1': {0: 1, 1: 0},\n",
       "             'C[C@@H](Oc1cc(Br)cnc1N(C(=O)OC(C)(C)C)C(=O)OC(C)(C)C)c1c(Cl)ccc(F)c1Cl': {0: 1,\n",
       "              1: 0},\n",
       "             'CC(C)(C)OC(=O)N1CCC(n2cc(B3OC(C)(C)C(C)(C)O3)cn2)CC1': {0: 16,\n",
       "              1: 5},\n",
       "             'C[C@@H](Oc1cc(-c2cnn(C3CCN(C(=O)OC(C)(C)C)CC3)c2)cnc1N(C(=O)OC(C)(C)C)C(=O)OC(C)(C)C)c1c(Cl)ccc(F)c1Cl': {0: 1,\n",
       "              1: 0},\n",
       "             'Brc1cc(C=Cc2c3nc(c(-c4ccccc4)c4ccc(cc5nc(c(-c6ccccc6)c6ccc2[nH]6)C=C5)[nH]4)C=C3)c(Br)s1': {0: 1,\n",
       "              1: 0},\n",
       "             'CCCCCCOc1csc(B(O)O)c1': {0: 2, 1: 0},\n",
       "             'CCCCCCOc1csc(-c2cc(/C=C/c3c4nc(c(-c5ccccc5)c5ccc(cc6nc(c(-c7ccccc7)c7ccc3[nH]7)C=C6)[nH]5)C=C4)c(-c3cc(OCCCCCC)cs3)s2)c1': {0: 1,\n",
       "              1: 0},\n",
       "             'N#Cc1cc(Br)ccc1N': {0: 3, 1: 0},\n",
       "             'N#Cc1cc(-c2ccccc2)ccc1N': {0: 1, 1: 0},\n",
       "             'CCCCOC(=O)c1sc2c(ccc3c(Cl)c(C(=O)OCCCC)sc32)c1Cl': {0: 9, 1: 0},\n",
       "             'CCCCOC(=O)c1sc2c(ccc3c(-c4ccc(OC)cc4)c(C(=O)OCCCC)sc32)c1-c1ccc(OC)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'O=C(Cc1c[nH]c2ccc(Br)cc12)NC12CC3CC(CC(C3)C1)C2': {0: 1, 1: 0},\n",
       "             'O=C(Cc1c[nH]c2ccc(-c3ccccc3)cc12)NC12CC3CC(CC(C3)C1)C2': {0: 1,\n",
       "              1: 0},\n",
       "             'C=C(CBr)c1cccc(OC)c1': {0: 6, 1: 0},\n",
       "             'C=C(Cc1cccc(OC)c1)c1cccc(OC)c1': {0: 2, 1: 0},\n",
       "             'Nc1nccc(Cl)n1': {0: 4, 1: 0},\n",
       "             'OB(O)c1ccc2c(c1)OCO2': {0: 222, 1: 8},\n",
       "             'Nc1nccc(-c2ccc3c(c2)OCO3)n1': {0: 1, 1: 0},\n",
       "             'CNc1nc2ccccc2n2c(Br)cnc12': {0: 11, 1: 1},\n",
       "             'CNc1nc2ccccc2n2c(-c3ccc(OC)cc3)cnc12': {0: 1, 1: 0},\n",
       "             'Brc1cnc2ccccc2c1': {0: 70, 1: 1},\n",
       "             'F[B-](F)(F)C1CCCC1': {0: 27, 1: 0},\n",
       "             'c1ccc2ncc(C3CCCC3)cc2c1': {0: 1, 1: 0},\n",
       "             'COC1=C(Br)C(=O)C(OC)=C(c2ccc(OC)cc2)C1=O': {0: 9, 1: 0},\n",
       "             'COC1=C(c2ccc(OC)cc2)C(=O)C(OC)=C(c2ccccc2OC)C1=O': {0: 1, 1: 0},\n",
       "             'CC(C)(C)OC(=O)N1CCC2(C=C(OS(=O)(=O)C(F)(F)F)c3ccccc3O2)CC1': {0: 4,\n",
       "              1: 0},\n",
       "             'CC(C)(C)OC(=O)N1CCC2(C=C(c3ccc(C#N)cc3)c3ccccc3O2)CC1': {0: 1,\n",
       "              1: 0},\n",
       "             'COC(=O)C1=Cc2cc(Br)ccc2S(=O)(=O)CC1': {0: 4, 1: 0},\n",
       "             'CC(C)c1ccc(B(O)O)cc1': {0: 28, 1: 3},\n",
       "             'COC(=O)C1=Cc2cc(-c3ccc(C(C)C)cc3)ccc2S(=O)(=O)CC1': {0: 1, 1: 0},\n",
       "             'O=C1c2ccccc2-c2[nH]nc(-c3ccc(Br)cc3)c21': {0: 2, 1: 0},\n",
       "             'Nc1cccc(B(O)O)c1': {0: 116, 1: 23},\n",
       "             'Nc1cccc(-c2ccc(-c3n[nH]c4c3C(=O)c3ccccc3-4)cc2)c1': {0: 1, 1: 0},\n",
       "             'CCCCCCOc1cc(Br)c(OCCCCCC)cc1Br': {0: 3, 1: 0},\n",
       "             'CC1(C)OB(c2ccc(/C=C/B3Nc4cccc5cccc(c45)N3)cc2)OC1(C)C': {0: 4,\n",
       "              1: 0},\n",
       "             'CCCCCCOc1cc(-c2ccc(/C=C/B3Nc4cccc5cccc(c45)N3)cc2)c(OCCCCCC)cc1-c1ccc(/C=C/B2Nc3cccc4cccc(c34)N2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'Nc1ccccc1Cl': {0: 5, 1: 1},\n",
       "             'Cc1ccccc1-c1ccccc1N': {0: 3, 1: 0},\n",
       "             'Oc1c(F)cc(Br)cc1F': {0: 4, 1: 1},\n",
       "             'OB(O)c1cc[nH]n1': {0: 5, 1: 11},\n",
       "             'Oc1c(F)cc(-c2cc[nH]n2)cc1F': {0: 1, 1: 0},\n",
       "             'Cc1cc(Cl)n(-c2ccccc2)n1': {0: 1, 1: 0},\n",
       "             'CC(C)=CCB1OC(C)(C)C(C)(C)O1': {0: 36, 1: 4},\n",
       "             'CC(C)=CCc1cc(C)nn1-c1ccccc1': {0: 1, 1: 0},\n",
       "             'COc1ccc(-c2ccc(OC)c(-c3ccccc3)c2)cc1Br': {0: 7, 1: 0},\n",
       "             'COc1ccc(-c2ccc(OC)c(-c3ccccc3)c2)cc1-c1ccccc1': {0: 2, 1: 0},\n",
       "             'Brc1ccccc1-c1nnc(-c2ccccc2Br)o1': {0: 2, 1: 0},\n",
       "             'OB(O)c1cccc(N(c2ccccc2)c2ccccc2)c1': {0: 2, 1: 0},\n",
       "             'c1ccc(N(c2ccccc2)c2cccc(-c3ccccc3-c3nnc(-c4ccccc4-c4cccc(N(c5ccccc5)c5ccccc5)c4)o3)c2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'COc1ccc(-c2ccc(OC(F)(F)F)cc2)cc1': {0: 5, 1: 0},\n",
       "             'Clc1nc2ccccc2n2cccc12': {0: 7, 1: 2},\n",
       "             'C=CB(OCCCC)OCCCC': {0: 9, 1: 2},\n",
       "             'C=Cc1nc2ccccc2n2cccc12': {0: 1, 1: 1},\n",
       "             'Cc1ccc(Br)cc1': {0: 191, 1: 9},\n",
       "             'COc1ccc(B2OCCO2)cc1C': {0: 4, 1: 0},\n",
       "             'COc1ccc(-c2ccc(C)cc2)cc1C': {0: 1, 1: 0},\n",
       "             'O=c1cc(N2CCOCC2)oc2cc(OS(=O)(=O)C(F)(F)F)ccc12': {0: 0, 1: 41},\n",
       "             'O=c1cc(N2CCOCC2)oc2cc(-c3ccsc3)ccc12': {0: 0, 1: 1},\n",
       "             'Ic1ccccc1': {0: 243, 1: 12},\n",
       "             'Cn1nc(B2OC(C)(C)C(C)(C)O2)c2ccccc21': {0: 10, 1: 1},\n",
       "             'Cn1nc(-c2ccccc2)c2ccccc21': {0: 5, 1: 0},\n",
       "             'CCOC(=O)c1cc(-c2ccc(C)cc2)ccc1-c1ccc(-c2ccc(-c3cc(OC)cc(OS(=O)(=O)C(F)(F)F)c3)c(F)c2)c(OC)c1': {0: 1,\n",
       "              1: 0},\n",
       "             'CC(=O)Nc1ccccc1B1OC(C)(C)C(C)(C)O1': {0: 2, 1: 3},\n",
       "             'CCOC(=O)c1cc(-c2ccc(C)cc2)ccc1-c1ccc(-c2ccc(-c3cc(OC)cc(-c4ccccc4NC(C)=O)c3)c(F)c2)c(OC)c1': {0: 1,\n",
       "              1: 0},\n",
       "             'Brc1cc(Br)c(Br)cc1Br': {0: 13, 1: 1},\n",
       "             'COc1ccc(-c2cc(-c3ccc(OC)cc3)c(-c3ccc(OC)cc3)cc2-c2ccc(OC)cc2)cc1': {0: 2,\n",
       "              1: 0},\n",
       "             'Cc1c(C)c(C)c(Br)c(C)c1C': {0: 1, 1: 0},\n",
       "             'CC(C)B(O)O': {0: 45, 1: 2},\n",
       "             'Cc1c(C)c(C)c(C(C)C)c(C)c1C': {0: 1, 1: 0},\n",
       "             'COc1cc(C#N)cc(Cl)c1OC': {0: 3, 1: 0},\n",
       "             'COc1cc(C#N)cc(C2CC2)c1OC': {0: 1, 1: 0},\n",
       "             'O=Cc1ccc(B(O)O)cc1': {0: 418, 1: 36},\n",
       "             'O=Cc1ccc(-c2ccc(C=O)cc2)cc1': {0: 1, 1: 0},\n",
       "             'O=[N+]([O-])c1ccc(O)c(I)c1': {0: 8, 1: 0},\n",
       "             '[2H]c1c([2H])c([2H])c(B(O)O)c([2H])c1[2H]': {0: 9, 1: 0},\n",
       "             '[2H]c1c([2H])c([2H])c(-c2cc([N+](=O)[O-])ccc2O)c([2H])c1[2H]': {0: 1,\n",
       "              1: 0},\n",
       "             'Cc1cccnc1Br': {0: 28, 1: 2},\n",
       "             'Cc1ccccc1-c1ncccc1C': {0: 2, 1: 0},\n",
       "             'C[Si](C)(C)CCOCn1cc(C(=O)Nc2ccc3[nH]ncc3c2)c2ncc(Br)cc21': {0: 1,\n",
       "              1: 0},\n",
       "             'O=C(Nc1ccc2[nH]ncc2c1)c1c[nH]c2cc(-c3ccccc3)cnc12': {0: 1, 1: 0},\n",
       "             'O=Cc1cc(-c2ccccc2)cnc1Cl': {0: 3, 1: 0},\n",
       "             'O=Cc1cc(-c2ccccc2)cnc1-c1ccc(OC(F)(F)F)cc1': {0: 1, 1: 0},\n",
       "             'Cc1cnc(Cl)c2c1c1ccccc1n2C': {0: 1, 1: 0},\n",
       "             'Cc1cnc(-c2ccccc2)c2c1c1ccccc1n2C': {0: 1, 1: 0},\n",
       "             'CCn1nc(-c2cccc(Br)c2)cc(NC(=O)Nc2c(Cl)cncc2Cl)c1=O': {0: 3,\n",
       "              1: 8},\n",
       "             'NC(=O)c1ccc(B(O)O)cc1': {0: 22, 1: 17},\n",
       "             'CCn1nc(-c2cccc(-c3ccc(C(N)=O)cc3)c2)cc(NC(=O)Nc2c(Cl)cncc2Cl)c1=O': {0: 0,\n",
       "              1: 1},\n",
       "             'O=Cc1ccccc1OS(=O)(=O)C(F)(F)F': {0: 2, 1: 0},\n",
       "             'COCOc1cccc(OCOC)c1B(O)O': {0: 8, 1: 0},\n",
       "             'COCOc1cccc(OCOC)c1-c1ccccc1C=O': {0: 1, 1: 0},\n",
       "             'Brc1c2ccccc2c(C(=C(c2ccccc2)c2ccccc2)c2ccccc2)c2ccccc12': {0: 1,\n",
       "              1: 0},\n",
       "             'c1ccc(C(=C(c2ccccc2)c2c3ccccc3c(-c3ccc(N(c4ccccc4)c4ccccc4)cc3)c3ccccc23)c2ccccc2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'Clc1cc(Cl)nc(Cl)n1': {0: 28, 1: 5},\n",
       "             'Cn1ccc(B(O)O)c1': {0: 1, 1: 0},\n",
       "             'Cn1ccc(-c2cc(Cl)nc(Cl)n2)c1': {0: 1, 1: 0},\n",
       "             'CC(=O)Nc1ncc(C)c(Br)n1': {0: 0, 1: 1},\n",
       "             'CCOC(=O)Cc1cn(Cc2csc3ccc(Cl)cc23)c2cc(B(O)O)ccc12': {0: 0, 1: 1},\n",
       "             'CCOC(=O)Cc1cn(Cc2csc3ccc(Cl)cc23)c2cc(-c3nc(N)ncc3C)ccc12': {0: 0,\n",
       "              1: 1},\n",
       "             'Brc1cnc(-c2ccc(N(c3ccccc3)c3ccccc3)cc2)c2nsnc12': {0: 2, 1: 0},\n",
       "             'N#Cc1ccc(-c2cnc(-c3ccc(N(c4ccccc4)c4ccccc4)cc3)c3nsnc23)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'Brc1cc2sc(Br)cc2s1': {0: 7, 1: 0},\n",
       "             'c1ccc(N(c2ccccc2)c2ccc(-c3cc4sc(-c5ccc(N(c6ccccc6)c6ccccc6)cc5)cc4s3)cc2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'C=C(CBr)c1ccc(C#N)cc1': {0: 3, 1: 0},\n",
       "             'C=C(Cc1ccccc1)c1ccc(C#N)cc1': {0: 1, 1: 0},\n",
       "             'CC(C)c1nn(-c2ccc(C#N)c(N[C@H]3CC[C@H](O)CC3)c2)c2nccc(I)c12': {0: 9,\n",
       "              1: 0},\n",
       "             'COc1cncc(B(O)O)c1': {0: 13, 1: 8},\n",
       "             'COc1cncc(-c2ccnc3c2c(C(C)C)nn3-c2ccc(C#N)c(N[C@H]3CC[C@H](O)CC3)c2)c1': {0: 1,\n",
       "              1: 0},\n",
       "             'Brc1ccccc1I': {0: 46, 1: 1},\n",
       "             'Cc1ccc(-c2ccccc2-c2ccc(C)cc2)cc1': {0: 3, 1: 0},\n",
       "             'Cc1cc(Br)ccc1C#N': {0: 6, 1: 0},\n",
       "             'Cc1cc(-c2ccccc2)ccc1C#N': {0: 1, 1: 0},\n",
       "             'CCOC(=O)c1[nH]c(I)c2ccc3nsnc3c12': {0: 2, 1: 0},\n",
       "             'CCOC(=O)c1[nH]c(-c2ccccc2)c2ccc3nsnc3c12': {0: 1, 1: 0},\n",
       "             'N#Cc1nc(-c2ccc(I)cc2)c(-c2ccc(I)cc2)nc1C#N': {0: 2, 1: 0},\n",
       "             'CN(C)c1ccc(C#Cc2ccc(B3OC(C)(C)C(C)(C)O3)cc2)cc1': {0: 5, 1: 0},\n",
       "             'CN(C)c1ccc(C#Cc2ccc(-c3ccc(-c4nc(C#N)c(C#N)nc4-c4ccc(-c5ccc(C#Cc6ccc(N(C)C)cc6)cc5)cc4)cc3)cc2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'CC(C)(C)[Si](C)(C)OCCC[n+]1c(-c2ccccc2)cc(-c2ccccc2)cc1-c1ccccc1': {0: 1,\n",
       "              1: 0},\n",
       "             'COc1ccc(B(O)O)cn1': {0: 105, 1: 12},\n",
       "             'COc1ccc(CCCO[Si](C)(C)C(C)(C)C)cn1': {0: 1, 1: 0},\n",
       "             'Brc1ccc2cc[nH]c2c1': {0: 13, 1: 5},\n",
       "             'Cc1noc(C)c1B(O)O': {0: 39, 1: 19},\n",
       "             'Cc1noc(C)c1-c1ccc2cc[nH]c2c1': {0: 0, 1: 1},\n",
       "             'COC(=O)c1c(OS(=O)(=O)C(F)(F)F)c(OS(=O)(=O)C(F)(F)F)c(C(=O)OC)n1Cc1ccccc1': {0: 10,\n",
       "              1: 0},\n",
       "             'C=Cc1c(C=C)c(C(=O)OC)n(Cc2ccccc2)c1C(=O)OC': {0: 1, 1: 0},\n",
       "             'COc1ccc(-c2ncc(-c3ccccc3)c(-c3ccccc3)c2C(=O)c2ccccc2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'N#Cc1ccc(Cl)cc1': {0: 50, 1: 2},\n",
       "             'OB(O)c1ccncc1': {0: 329, 1: 76},\n",
       "             'N#Cc1ccc(-c2ccncc2)cc1': {0: 3, 1: 0},\n",
       "             'COc1nc(Cl)nc2c1ncn2[C@H]1C[C@H](O[Si](C)(C)C(C)(C)C)[C@H](O[Si](C)(C)C(C)(C)C)[C@@H](CO[Si](C)(C)C(C)(C)C)O1': {0: 3,\n",
       "              1: 0},\n",
       "             'COc1nc(-c2ccccc2)nc2c1ncn2[C@H]1C[C@H](O[Si](C)(C)C(C)(C)C)[C@H](O[Si](C)(C)C(C)(C)C)[C@@H](CO[Si](C)(C)C(C)(C)C)O1': {0: 2,\n",
       "              1: 0},\n",
       "             'O=C1CCCc2c(Br)cccc21': {0: 2, 1: 1},\n",
       "             'O=C1CCCc2c1cccc2-c1ccccc1': {0: 1, 1: 0},\n",
       "             '[O-][n+]1c(Cl)c(OC2CCCC2)nc2ccc(F)cc21': {0: 7, 1: 1},\n",
       "             '[O-][n+]1c(-c2cccc3ccccc23)c(OC2CCCC2)nc2ccc(F)cc21': {0: 1,\n",
       "              1: 0},\n",
       "             'CC(C)(C)C(O)(c1cncnc1)c1ccc(Br)cc1C(F)(F)F': {0: 1, 1: 0},\n",
       "             'CC(C)(C)C(O)(c1cncnc1)c1ccc(-c2ccc(OC(F)(F)F)cc2)cc1C(F)(F)F': {0: 1,\n",
       "              1: 0},\n",
       "             'Cc1ccc2nc(Cl)nc(NCCN(C)C)c2c1': {0: 0, 1: 3},\n",
       "             'Cc1ccc2nc(-c3ccc4ccccc4c3)nc(NCCN(C)C)c2c1': {0: 0, 1: 1},\n",
       "             'Brc1ccc2[nH]c3ccc(Br)cc3c2c1': {0: 8, 1: 0},\n",
       "             'c1ccc(N(c2ccccc2)c2ccc(-c3ccc4[nH]c5ccc(-c6ccc(N(c7ccccc7)c7ccccc7)cc6)cc5c4c3)cc2)cc1': {0: 2,\n",
       "              1: 0},\n",
       "             'CCN(CC)C(=O)Oc1ccc(C#N)cc1': {0: 0, 1: 1},\n",
       "             'N#Cc1ccc(-c2ccccc2)cc1': {0: 28, 1: 2},\n",
       "             'COc1ccc(-c2cccc([N+](=O)[O-])c2)cc1': {0: 9, 1: 0},\n",
       "             'CC1(C)Cc2nc1cc1ccc(cc3nc(cc4ccc([nH]4)c2Br)C=C3)[nH]1': {0: 1,\n",
       "              1: 1},\n",
       "             'COC(=O)c1cc(B2OC(C)(C)C(C)(C)O2)cc(C(=O)OC)n1': {0: 0, 1: 1},\n",
       "             'COC(=O)c1cc(-c2c3nc(cc4ccc(cc5nc(cc6ccc2[nH]6)C=C5)[nH]4)C(C)(C)C3)cc(C(=O)OC)n1': {0: 0,\n",
       "              1: 1},\n",
       "             'COc1cc2c(cc1Br)Cc1cc(OC)c(Br)cc1Cc1cc(OC)c(Br)cc1C2': {0: 2,\n",
       "              1: 3},\n",
       "             'COc1cc2c(cc1-c1ccncc1)Cc1cc(OC)c(-c3ccncc3)cc1Cc1cc(OC)c(-c3ccncc3)cc1C2': {0: 0,\n",
       "              1: 1},\n",
       "             'CCCCC(CC)Cn1c2ccc3cccc4c5cccc6c(Br)cc1c(c65)c2c34': {0: 2, 1: 0},\n",
       "             'COc1ccc(N(c2ccc(OC)cc2)c2ccc(B(O)O)cc2)cc1': {0: 14, 1: 2},\n",
       "             'CCCCC(CC)Cn1c2ccc3cccc4c5cccc6c(-c7ccc(N(c8ccc(OC)cc8)c8ccc(OC)cc8)cc7)cc1c(c65)c2c34': {0: 1,\n",
       "              1: 0},\n",
       "             'O=S(=O)(OC1=CCCCC1)C(F)(F)F': {0: 9, 1: 1},\n",
       "             'OB(O)c1c(F)cccc1F': {0: 21, 1: 2},\n",
       "             'Fc1cccc(F)c1C1=CCCCC1': {0: 1, 1: 0},\n",
       "             'C[C@@H](Nc1ncnc2sc(Br)cc12)c1ccccc1': {0: 20, 1: 4},\n",
       "             'OCc1cccc(B(O)O)c1': {0: 58, 1: 15},\n",
       "             'C[C@@H](Nc1ncnc2sc(-c3cccc(CO)c3)cc12)c1ccccc1': {0: 1, 1: 0},\n",
       "             'BrCc1ccccc1': {0: 62, 1: 0},\n",
       "             'c1ccc(Cc2cccs2)cc1': {0: 4, 1: 0},\n",
       "             'CCOP(=O)(OCC)OC[C@H]1O[C@@H](n2cnc3c(Cl)ncnc32)[C@@H]2OC(C)(C)O[C@H]12': {0: 7,\n",
       "              1: 0},\n",
       "             'O=C(O)c1ccc(B(O)O)cc1': {0: 84, 1: 15},\n",
       "             'CCOP(=O)(OCC)OC[C@H]1O[C@@H](n2cnc3c(-c4ccc(C(=O)O)cc4)ncnc32)[C@@H]2OC(C)(C)O[C@H]12': {0: 1,\n",
       "              1: 0},\n",
       "             'CC(C)(C)OC(=O)N1[C@H]2CC[C@@H]1C=C(OS(=O)(=O)C(F)(F)F)C2': {0: 1,\n",
       "              1: 0},\n",
       "             'CC1(C)OB(B2OC(C)(C)C(C)(C)O2)OC1(C)C': {0: 398, 1: 19},\n",
       "             'CC(C)(C)OC(=O)N1[C@H]2CC[C@@H]1C=C(B1OC(C)(C)C(C)(C)O1)C2': {0: 1,\n",
       "              1: 0},\n",
       "             'Brc1ccc2ccccc2n1': {0: 23, 1: 3},\n",
       "             'c1ccc(N(c2ccccc2)c2ccc(-c3ccc4ccccc4n3)cc2)cc1': {0: 1, 1: 0},\n",
       "             'CS(=O)(=O)OCc1ccccc1': {0: 4, 1: 0},\n",
       "             'COc1ccc(Cc2ccccc2)cc1': {0: 21, 1: 0},\n",
       "             'O=C(Oc1ccccc1)c1ccccc1': {0: 31, 1: 2},\n",
       "             'O=C(c1ccccc1)c1ccc(F)cc1': {0: 23, 1: 0},\n",
       "             'CC(=O)c1ccc(Cl)cc1': {0: 72, 1: 2},\n",
       "             'F[B-](F)(F)CN1CCCCC1': {0: 32, 1: 1},\n",
       "             'CC(=O)c1ccc(CN2CCCCC2)cc1': {0: 2, 1: 1},\n",
       "             'COC(=O)c1sccc1Br': {0: 15, 1: 1},\n",
       "             'COC(=O)c1sccc1-c1cc(OC)c(OC)c(OC)c1': {0: 1, 1: 0},\n",
       "             'Cc1ccc(S(=O)(=O)OC2=CC(=O)N(C(=O)OC(C)(C)C)[C@H]2Cc2ccccc2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'CC(C)(C)OC(=O)N1C(=O)C=C(c2ccccc2)[C@@H]1Cc1ccccc1': {0: 1,\n",
       "              1: 0},\n",
       "             'Cc1ccc(-c2ccccc2I)cc1': {0: 2, 1: 0},\n",
       "             'Cc1ccc(-c2ccccc2-c2ccccc2)cc1': {0: 7, 1: 0},\n",
       "             'COc1ccc(CN2CC(OS(=O)(=O)C(F)(F)F)=C(c3ccccc3)C2=O)cc1OC': {0: 5,\n",
       "              1: 0},\n",
       "             'COc1ccc(C2=C(c3ccccc3)C(=O)N(Cc3ccc(OC)c(OC)c3)C2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'O=C(c1ccccc1)n1cccc1': {0: 6, 1: 0},\n",
       "             'COc1ccccc1C(=O)c1ccccc1': {0: 8, 1: 0},\n",
       "             'Cc1nn(-c2c(F)cccc2F)c2c1cc(Br)c(=O)n2C': {0: 0, 1: 1},\n",
       "             'Cc1ccc(C(=O)NC2CC2)cc1B1OC(C)(C)C(C)(C)O1': {0: 7, 1: 6},\n",
       "             'Cc1ccc(C(=O)NC2CC2)cc1-c1cc2c(C)nn(-c3c(F)cccc3F)c2n(C)c1=O': {0: 0,\n",
       "              1: 1},\n",
       "             'O=C1CNC(=O)/C(=C/c2c[nH]c3cc(Br)ccc23)N1': {0: 2, 1: 0},\n",
       "             'O=C1CNC(=O)/C(=C/c2c[nH]c3cc(-c4ccccc4)ccc23)N1': {0: 1, 1: 0},\n",
       "             'BrCCC1OCCCO1': {0: 2, 1: 0},\n",
       "             'CCCCCC/C(=C\\\\B1OC(C)(C)C(C)(C)O1)CC': {0: 8, 1: 0},\n",
       "             'CCCCCC/C(=C\\\\CCC1OCCCO1)CC': {0: 1, 1: 0},\n",
       "             'Cc1nc(Cl)cc(Cl)n1': {0: 9, 1: 1},\n",
       "             'COc1ccc(-c2cc(-c3ccc(OC)cc3)nc(C)n2)cc1': {0: 2, 1: 0},\n",
       "             'CC(C)(C)OC(=O)NCCOc1ccc(Br)cc1C=O': {0: 1, 1: 0},\n",
       "             'OB(O)c1csc2ccccc12': {0: 40, 1: 6},\n",
       "             'CC(C)(C)OC(=O)NCCOc1ccc(-c2csc3ccccc23)cc1C=O': {0: 1, 1: 0},\n",
       "             'COc1ccc(I)cc1OCc1ccccc1': {0: 3, 1: 0},\n",
       "             'OB(O)c1ccccc1Br': {0: 99, 1: 6},\n",
       "             'COc1ccc(-c2ccccc2Br)cc1OCc1ccccc1': {0: 1, 1: 0},\n",
       "             'COc1nc(Cl)c(C=O)c(OC)n1': {0: 1, 1: 0},\n",
       "             'C=C(B1OC(C)(C)C(C)(C)O1)c1ccc(F)cc1': {0: 1, 1: 0},\n",
       "             'C=C(c1ccc(F)cc1)c1nc(OC)nc(OC)c1C=O': {0: 1, 1: 0},\n",
       "             'O=S(=O)(F)c1cncc(Br)c1': {0: 4, 1: 0},\n",
       "             'OB(O)C1CC1': {0: 129, 1: 31},\n",
       "             'O=S(=O)(F)c1cncc(C2CC2)c1': {0: 1, 1: 0},\n",
       "             'Cc1c(-c2ccc(Br)s2)sc(-c2ccc(Br)s2)c1C': {0: 1, 1: 0},\n",
       "             'Cc1c(-c2ccc(-c3cccs3)s2)sc(-c2ccc(-c3cccs3)s2)c1C': {0: 1, 1: 0},\n",
       "             'C[Si](C)(C)c1ccc(B(O)O)cc1': {0: 36, 1: 1},\n",
       "             'COC(=O)c1cc(-c2ccc([Si](C)(C)C)cc2)c(C(=O)OC)cc1-c1ccc([Si](C)(C)C)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'C=C(OS(=O)(=O)C(F)(F)C(F)(F)C(F)(F)C(F)(F)F)P(=O)(OCC)OCC': {0: 25,\n",
       "              1: 0},\n",
       "             'C=C(c1ccc2ccccc2c1)P(=O)(OCC)OCC': {0: 2, 1: 0},\n",
       "             'Cc1cccc([B-](O)(O)O)c1': {0: 4, 1: 0},\n",
       "             'Cc1ccc(-c2cccc(C)c2)cc1': {0: 10, 1: 0},\n",
       "             'CN(C)C(=O)OCc1ccccc1': {0: 15, 1: 0},\n",
       "             'c1ccc(Cc2ccc(-c3ccccc3)cc2)cc1': {0: 8, 1: 1},\n",
       "             'O=[N+]([O-])c1cccc(Cl)c1': {0: 7, 1: 0},\n",
       "             'CC(C)O[B-](OC(C)C)(OC(C)C)c1cnn(C)c1': {0: 8, 1: 0},\n",
       "             'Cn1cc(-c2cccc([N+](=O)[O-])c2)cn1': {0: 1, 1: 0},\n",
       "             'Brc1ccnc(-c2ccccc2)c1': {0: 1, 1: 0},\n",
       "             'Cc1ccc(-c2ccnc(-c3ccccc3)c2)cc1': {0: 1, 1: 0},\n",
       "             'O=[N+]([O-])c1ccc(Br)nc1': {0: 19, 1: 1},\n",
       "             'CC(C)(C)OC(=O)N1CC=C(B2OC(C)(C)C(C)(C)O2)CC1': {0: 113, 1: 10},\n",
       "             'CC(C)(C)OC(=O)N1CC=C(c2ccc([N+](=O)[O-])cn2)CC1': {0: 1, 1: 0},\n",
       "             'COc1ccc(-c2cc(OS(=O)(=O)C(F)(F)F)c3cccn3c2)cc1': {0: 1, 1: 0},\n",
       "             'COc1ccc(-c2cc(-c3ccc([N+](=O)[O-])cc3)c3cccn3c2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'CC(C)(c1cccc2ccccc12)S(=O)(=O)c1ccccc1': {0: 1, 1: 0},\n",
       "             'COc1ccc(B2OB(c3ccc(OC)cc3)OB(c3ccc(OC)cc3)O2)cc1': {0: 18, 1: 1},\n",
       "             'COc1ccc(C(C)(C)c2cccc3ccccc23)cc1': {0: 1, 1: 0},\n",
       "             'COc1ccc(C(=O)/C=C/c2ccccc2)cc1OC': {0: 3, 1: 0},\n",
       "             'OB(O)c1ccc(F)nc1': {0: 82, 1: 3},\n",
       "             'COc1ccc(C(=O)/C=C/c2ccccc2-c2ccc(F)nc2)cc1OC': {0: 1, 1: 0},\n",
       "             'O=C(CCl)OCc1ccccc1': {0: 13, 1: 0},\n",
       "             'N#CCCC/C=C/[B-](F)(F)F': {0: 1, 1: 0},\n",
       "             'N#CCCC/C=C/CC(=O)OCc1ccccc1': {0: 1, 1: 0},\n",
       "             'CC(=O)Nc1c(Br)c(-c2ccccc2)nn1C': {0: 10, 1: 0},\n",
       "             'CC(=O)Nc1c(-c2ccccc2C)c(-c2ccccc2)nn1C': {0: 1, 1: 0},\n",
       "             'COC(=O)c1cn2ccccc2c1C(=O)OC': {0: 3, 1: 0},\n",
       "             'COC(=O)c1c(C(=O)OC)c2ccccn2c1-c1ccc(C(F)(F)F)cc1': {0: 1, 1: 0},\n",
       "             'CC(=O)OC[C@H]1O[C@@H](n2cnc3c(Cl)nc(COC(=O)c4ccccc4)nc32)[C@H](OC(C)=O)[C@@H]1OC(C)=O': {0: 2,\n",
       "              1: 0},\n",
       "             'CC(=O)OC[C@H]1O[C@@H](n2cnc3c(-c4ccccc4)nc(COC(=O)c4ccccc4)nc32)[C@H](OC(C)=O)[C@@H]1OC(C)=O': {0: 1,\n",
       "              1: 0},\n",
       "             'COC(=O)C(/C=C/c1ccccc1)(C(=O)OC(C)(C)C)n1cc(Br)ccc1=O': {0: 1,\n",
       "              1: 0},\n",
       "             'COC(=O)C(/C=C/c1ccccc1)(C(=O)OC(C)(C)C)n1cc(-c2ccccc2)ccc1=O': {0: 1,\n",
       "              1: 0},\n",
       "             'COc1ccc(Cn2cc(Br)c[n+]2[O-])cc1': {0: 2, 1: 0},\n",
       "             'COc1ccc(Cn2cc(-c3ccccc3C=O)c[n+]2[O-])cc1': {0: 1, 1: 0},\n",
       "             'CC(C)[C@@H]1CC[C@@H](C)C[C@H]1[P@]1(=O)c2ccccc2-c2cc(OS(=O)(=O)C(F)(F)F)ccc21': {0: 1,\n",
       "              1: 0},\n",
       "             'CCC/C(B1OC(C)(C)C(C)(C)O1)=C(\\\\CCC)c1ccc2ccc3ccc4ccccc4c3c2c1': {0: 5,\n",
       "              1: 0},\n",
       "             'CCC/C(=C(\\\\CCC)c1ccc2ccc3ccc4ccccc4c3c2c1)c1ccc2c(c1)-c1ccccc1[P@@]2(=O)[C@@H]1C[C@H](C)CC[C@H]1C(C)C': {0: 1,\n",
       "              1: 0},\n",
       "             'O=c1[nH]c2ncc(Br)nc2n1C1CCCCC1': {0: 0, 1: 8},\n",
       "             'O=c1[nH]c2ncc(-c3cccnc3)nc2n1C1CCCCC1': {0: 0, 1: 1},\n",
       "             'CSc1nn2c(I)cccc2c1NC(=O)OC(C)(C)C': {0: 1, 1: 0},\n",
       "             'COc1cc(C)cc(OC)c1B(O)O': {0: 3, 1: 0},\n",
       "             'COc1cc(C)cc(OC)c1-c1cccc2c(NC(=O)OC(C)(C)C)c(SC)nn12': {0: 1,\n",
       "              1: 0},\n",
       "             'CC(=O)c1ccc(OS(=O)(=O)N(C)C)cc1': {0: 9, 1: 0},\n",
       "             'CC(C)(C)OC(=O)NC[B-](F)(F)F': {0: 54, 1: 3},\n",
       "             'CC(=O)c1ccc(CNC(=O)OC(C)(C)C)cc1': {0: 3, 1: 0},\n",
       "             'CCc1nc(N)nc(N)c1I': {0: 1, 1: 0},\n",
       "             'CCc1nc(N)nc(N)c1-c1ccc(Cl)cc1': {0: 1, 1: 0},\n",
       "             'Cn1cc(Br)cn1': {0: 2, 1: 2},\n",
       "             'CC1(C)OB(c2cc([Si](C)(C)C)sc2Cl)OC1(C)C': {0: 5, 1: 0},\n",
       "             'Cn1cc(-c2cc([Si](C)(C)C)sc2Cl)cn1': {0: 2, 1: 0},\n",
       "             'Fc1cccc2c1cc1c3nc(Cl)ccc3ncn21': {0: 0, 1: 3},\n",
       "             'CNC(=O)c1c(-n2ccccc2=O)oc2cc(N(C)S(C)(=O)=O)c(B3OC(C)(C)C(C)(C)O3)cc12': {0: 0,\n",
       "              1: 1},\n",
       "             'CNC(=O)c1c(-n2ccccc2=O)oc2cc(N(C)S(C)(=O)=O)c(-c3ccc4ncn5c6cccc(F)c6cc5c4n3)cc12': {0: 0,\n",
       "              1: 1},\n",
       "             'CCCCCCCCOc1c(OCCCCCCCC)c(-c2ccc(Br)s2)c2nsnc2c1-c1ccc(Br)s1': {0: 2,\n",
       "              1: 0},\n",
       "             'CCCCCCCCOc1c(OCCCCCCCC)c(-c2ccc(-c3ccc(N(c4ccccc4)c4ccccc4)cc3)s2)c2nsnc2c1-c1ccc(-c2ccc(N(c3ccccc3)c3ccccc3)cc2)s1': {0: 1,\n",
       "              1: 0},\n",
       "             'CC1(C)OB(c2cccc(CN3C(=O)c4ccccc4C3=O)c2)OC1(C)C': {0: 0, 1: 3},\n",
       "             'O=C1c2ccccc2C(=O)N1Cc1cccc(-c2ccc3c(=O)cc(N4CCOCC4)oc3c2)c1': {0: 0,\n",
       "              1: 1},\n",
       "             'Cc1cc2nc(N3CCNCC3)n(Cc3ccc(F)c(Cl)c3)c2cc1C': {0: 0, 1: 1},\n",
       "             'Cc1cc2nc(N3CCNCC3)n(Cc3ccc(F)c(C4CC4)c3)c2cc1C': {0: 0, 1: 1},\n",
       "             'Cc1cccc(C)c1Cl': {0: 53, 1: 0},\n",
       "             'CC(C)(C)c1ccc(B(O)O)cc1': {0: 392, 1: 12},\n",
       "             'Cc1cccc(C)c1-c1ccc(C(C)(C)C)cc1': {0: 5, 1: 0},\n",
       "             'CC(C)(C)OC(=O)N1CCC(C)(Oc2cc(Br)cc3c2cnn3C2CCCCO2)CC1': {0: 1,\n",
       "              1: 0},\n",
       "             'CC(C)(C)OC(=O)N1CCC(C)(Oc2cc(-c3ccc(O)cc3)cc3c2cnn3C2CCCCO2)CC1': {0: 1,\n",
       "              1: 0},\n",
       "             'CN(C)C(=O)OCc1ccc(F)cc1': {0: 7, 1: 0},\n",
       "             'COc1cccc(Cc2ccc(F)cc2)c1': {0: 2, 1: 0},\n",
       "             'C=CCN1CCc2cc(Br)ccc2C1=O': {0: 2, 1: 0},\n",
       "             'C=CCN1CCc2cc(-c3ccc(C#N)cc3)ccc2C1=O': {0: 1, 1: 0},\n",
       "             'COc1cccc(OC)c1C(=O)O': {0: 4, 1: 0},\n",
       "             'CC1(C)COB(c2ccc3c(c2)OCO3)OC1': {0: 5, 1: 0},\n",
       "             'COc1cccc(OC)c1-c1ccc2c(c1)OCO2': {0: 1, 1: 0},\n",
       "             'OC12CC3CC(C1)CC(NCc1cc(Br)on1)(C3)C2': {0: 1, 1: 0},\n",
       "             'CSc1ccccc1-c1cc(CNC23CC4CC(CC(O)(C4)C2)C3)no1': {0: 1, 1: 0},\n",
       "             'Cc1c(I)c(C)c(I)c(C)c1I': {0: 3, 1: 0},\n",
       "             'Cc1c(B2OC(C)(C)C(C)(C)O2)c(C)c(B2OC(C)(C)C(C)(C)O2)c(C)c1B1OC(C)(C)C(C)(C)O1': {0: 1,\n",
       "              1: 0},\n",
       "             'N#Cc1ccc(-c2cccc3ccccc23)cc1': {0: 5, 1: 1},\n",
       "             'Cc1cc(-c2ccccc2C=C(Br)Br)cs1': {0: 1, 1: 0},\n",
       "             'CC(=O)c1ccc(-c2cc3ccccc3c3cc(C)sc23)cc1': {0: 1, 1: 0},\n",
       "             'CCCCC1=C(Br)C(C)(C)OP1(=O)OCC': {0: 7, 1: 0},\n",
       "             'CCCCC1=C(c2ccc3c(c2)OCO3)C(C)(C)OP1(=O)OCC': {0: 1, 1: 0},\n",
       "             'COc1cc2c(c(OC)c1OC)C(=O)/C(=C/Br)O2': {0: 3, 1: 0},\n",
       "             'COc1ccc(/C=C2\\\\Oc3cc(OC)c(OC)c(OC)c3C2=O)cc1': {0: 1, 1: 0},\n",
       "             'COc1cc2nc3c(cnc4ccc(Br)cc43)cc2c(OC)c1OC': {0: 1, 1: 0},\n",
       "             'COc1cc2nc3c(cnc4ccc(-c5ccc(C=O)cc5)cc43)cc2c(OC)c1OC': {0: 1,\n",
       "              1: 0},\n",
       "             'Brc1cc2cc3c(cc2cc1Br)C1C=CC3c2cc3cc(Br)c(Br)cc3cc21': {0: 1,\n",
       "              1: 0},\n",
       "             'CC(C)(C)c1ccc(-c2cc3cc4c(cc3cc2-c2ccc(C(C)(C)C)cc2)C2C=CC4c3cc4cc(-c5ccc(C(C)(C)C)cc5)c(-c5ccc(C(C)(C)C)cc5)cc4cc32)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'CCOC(=O)c1nc(Br)n(C)c1-c1ccccc1': {0: 5, 1: 0},\n",
       "             'CCOC(=O)c1nc(-c2cccc3ccccc23)n(C)c1-c1ccccc1': {0: 1, 1: 0},\n",
       "             'O=S(=O)(Oc1ccc(Br)c2c1oc1ccccc12)C(F)(F)F': {0: 1, 1: 0},\n",
       "             'OB(O)c1ccc2c(c1)c1ccccc1n2-c1ccccc1': {0: 60, 1: 1},\n",
       "             'Brc1ccc(-c2ccc3c(c2)c2ccccc2n3-c2ccccc2)c2oc3ccccc3c12': {0: 1,\n",
       "              1: 0},\n",
       "             'COC[C@@H]1CCCN1C(=O)c1cc(I)ccc1OC': {0: 5, 1: 0},\n",
       "             'COC[C@@H]1CCCN1C(=O)c1cc(-c2ccccc2)ccc1OC': {0: 2, 1: 0},\n",
       "             'CC(=C\\\\C(Br)=C\\\\[C@@H](C)[C@@H](/C=C/C(Br)=C/CCc1ccccc1)O[Si](C)(C)C(C)(C)C)/C=C(\\\\C)C(=O)OCC[Si](C)(C)C': {0: 1,\n",
       "              1: 0},\n",
       "             'CB1C2CCCC1CCC2': {0: 1, 1: 0},\n",
       "             'CC(=C\\\\C(C)=C\\\\[C@@H](C)[C@@H](/C=C/C(Br)=C/CCc1ccccc1)O[Si](C)(C)C(C)(C)C)/C=C(\\\\C)C(=O)OCC[Si](C)(C)C': {0: 1,\n",
       "              1: 0},\n",
       "             'Clc1cc(Cl)ncn1': {0: 29, 1: 7},\n",
       "             'Fc1ccc(-c2cc(-c3ccc(F)cc3)ncn2)cc1': {0: 1, 1: 0},\n",
       "             'O=c1nc(N2CCOCC2)oc2c(Br)c(OCc3ccccn3)ccc12': {0: 0, 1: 5},\n",
       "             'COc1ccc(-c2c(OCc3ccccn3)ccc3c(=O)nc(N4CCOCC4)oc23)cc1': {0: 0,\n",
       "              1: 1},\n",
       "             'Brc1ccc(-c2ccccn2)cc1': {0: 9, 1: 1},\n",
       "             'CCCCCCCCC1(CCCCCCCC)c2cc(B3OC(C)(C)C(C)(C)O3)ccc2-c2ccc([Si](C)(C)C)cc21': {0: 8,\n",
       "              1: 0},\n",
       "             'CCCCCCCCC1(CCCCCCCC)c2cc(-c3ccc(-c4ccccn4)cc3)ccc2-c2ccc([Si](C)(C)C)cc21': {0: 1,\n",
       "              1: 0},\n",
       "             'Cc1nc2cc(Cl)ccc2s1': {0: 7, 1: 0},\n",
       "             'Cc1ccc(-c2ccc3sc(C)nc3c2)cc1': {0: 1, 1: 0},\n",
       "             'CC(O)c1cncc(Br)c1': {0: 1, 1: 0},\n",
       "             'Cc1nnc2n1-c1c(F)cc(B(O)O)cc1CC2': {0: 10, 1: 0},\n",
       "             'Cc1nnc2n1-c1c(F)cc(-c3cncc(C(C)O)c3)cc1CC2': {0: 1, 1: 0},\n",
       "             'Cc1ccc(I)cc1': {0: 92, 1: 5},\n",
       "             'CC1(C)OB(c2ccsc2-c2ccccc2)OC1(C)C': {0: 1, 1: 0},\n",
       "             'Cc1ccc(-c2ccsc2-c2ccccc2)cc1': {0: 1, 1: 0},\n",
       "             'Cc1cc(=O)oc2cc(OCc3cn(CCCNc4nc(Br)cn5ccnc45)nn3)ccc12': {0: 7,\n",
       "              1: 1},\n",
       "             'Cc1cc(=O)oc2cc(OCc3cn(CCCNc4nc(-c5ccsc5)cn5ccnc45)nn3)ccc12': {0: 0,\n",
       "              1: 1},\n",
       "             'Cc1ccc(Br)cc1S(=O)(=O)NC1CCCC1': {0: 3, 1: 3},\n",
       "             'Cc1noc(CO)c1B1OC(C)(C)C(C)(C)O1': {0: 0, 1: 1},\n",
       "             'Cc1ccc(-c2c(C)noc2CO)cc1S(=O)(=O)NC1CCCC1': {0: 0, 1: 1},\n",
       "             'COc1cc(Br)c2ccccc2c1O[Si](C)(C)C(C)(C)C': {0: 1, 1: 0},\n",
       "             'COc1cc(-c2ccccc2)c2ccccc2c1O[Si](C)(C)C(C)(C)C': {0: 1, 1: 0},\n",
       "             'CCCCCCCCCCCCCCCCCCOc1cc(COC(=O)c2ccc(I)cc2)cc(OCCCCCCCCCCCCCCCCCC)c1OCCCCCCCCCCCCCCCCCC': {0: 1,\n",
       "              1: 0},\n",
       "             'CCCCCCCCCCCCCCCCCCOc1cc(CSc2ccc(B(O)O)cc2)cc(OCCCCCCCCCCCCCCCCCC)c1OCCCCCCCCCCCCCCCCCC': {0: 6,\n",
       "              1: 0},\n",
       "             'CCCCCCCCCCCCCCCCCCOc1cc(COC(=O)c2ccc(-c3ccc(SCc4cc(OCCCCCCCCCCCCCCCCCC)c(OCCCCCCCCCCCCCCCCCC)c(OCCCCCCCCCCCCCCCCCC)c4)cc3)cc2)cc(OCCCCCCCCCCCCCCCCCC)c1OCCCCCCCCCCCCCCCCCC': {0: 1,\n",
       "              1: 0},\n",
       "             'CC1(c2ccccc2)CCN(c2cccc(Br)c2)C(=O)O1': {0: 0, 1: 5},\n",
       "             'CC1(c2ccccc2)CCN(c2cccc(-c3cccc(F)c3)c2)C(=O)O1': {0: 0, 1: 1},\n",
       "             '[O-][n+]1cc(Br)cc2ccccc21': {0: 7, 1: 0},\n",
       "             '[O-][n+]1cc(-c2ccccc2)cc2ccccc21': {0: 1, 1: 0},\n",
       "             'CC1(C)C=Cc2c(ccc([N+](=O)[O-])c2OS(=O)(=O)C(F)(F)F)O1': {0: 24,\n",
       "              1: 0},\n",
       "             'COc1ccc(B(O)O)cc1C': {0: 29, 1: 2},\n",
       "             'COc1ccc(-c2c([N+](=O)[O-])ccc3c2C=CC(C)(C)O3)cc1C': {0: 1, 1: 0},\n",
       "             'Brc1ccc(Br)s1': {0: 27, 1: 7},\n",
       "             'CCCCCCc1ccc(-c2ccc(B3OC(C)(C)C(C)(C)O3)s2)s1': {0: 28, 1: 7},\n",
       "             'CCCCCCc1ccc(-c2ccc(-c3ccc(-c4ccc(-c5ccc(CCCCCC)s5)s4)s3)s2)s1': {0: 1,\n",
       "              1: 1},\n",
       "             'Brc1c2c(nc3ccccc13)CCCC2': {0: 1, 1: 0},\n",
       "             'Cc1ccc(Nc2c3c(nc4ccccc24)CCCC3)cc1': {0: 1, 1: 0},\n",
       "             'O=[N+]([O-])c1ccc(C#CI)cc1': {0: 2, 1: 0},\n",
       "             'O=[N+]([O-])c1ccc(C#Cc2ccccc2)cc1': {0: 4, 1: 0},\n",
       "             'Oc1ccc(I)cc1': {0: 26, 1: 2},\n",
       "             'Cc1ccccc1-c1ccc(O)cc1': {0: 2, 1: 0},\n",
       "             'C=C(CBr)c1ccccc1': {0: 28, 1: 0},\n",
       "             'C=C(Cc1cccs1)c1ccccc1': {0: 1, 1: 0},\n",
       "             'Brc1ccc(-c2ccc(-c3cccs3)s2)s1': {0: 5, 1: 0},\n",
       "             'CSc1ccc(-c2ccc(-c3ccc(-c4cccs4)s3)s2)cc1': {0: 1, 1: 0},\n",
       "             'N#Cc1c(Br)[nH]c2nc(N)[nH]c(=O)c12': {0: 2, 1: 3},\n",
       "             'CS(=O)(=O)Nc1cccc(B(O)O)c1': {0: 14, 1: 9},\n",
       "             'CS(=O)(=O)Nc1cccc(-c2[nH]c3nc(N)[nH]c(=O)c3c2C#N)c1': {0: 0,\n",
       "              1: 1},\n",
       "             'Ic1csc(-c2ccccc2)n1': {0: 4, 1: 0},\n",
       "             'c1ccc(-c2csc(-c3ccccc3)n2)cc1': {0: 4, 1: 0},\n",
       "             'CC1(C)COB(c2ccc(C=O)cc2)OC1': {0: 2, 1: 0},\n",
       "             'O=Cc1ccc(-c2ccc([N+](=O)[O-])cc2)cc1': {0: 4, 1: 0},\n",
       "             'CN1C(=O)CCc2cc(Br)cc(F)c21': {0: 1, 1: 0},\n",
       "             'CN1C(=O)CCc2cc(B3OC(C)(C)C(C)(C)O3)cc(F)c21': {0: 1, 1: 0},\n",
       "             'Brc1csc(Br)n1': {0: 16, 1: 1},\n",
       "             'Brc1csc(C2CC2)n1': {0: 1, 1: 0},\n",
       "             'BrC1=Cc2c(ccc3cccc(-c4ccccc4)c23)OC1(c1ccccc1)c1ccccc1': {0: 1,\n",
       "              1: 0},\n",
       "             'C1=C(c2ccccc2)C(c2ccccc2)(c2ccccc2)Oc2ccc3cccc(-c4ccccc4)c3c21': {0: 1,\n",
       "              1: 0},\n",
       "             'Nc1ncc(Br)cc1[N+](=O)[O-]': {0: 6, 1: 0},\n",
       "             'COc1ccc(-n2nc(C(F)(F)F)cc2B2OC(C)(C)C(C)(C)O2)cc1': {0: 4, 1: 0},\n",
       "             'COc1ccc(-n2nc(C(F)(F)F)cc2-c2cnc(N)c([N+](=O)[O-])c2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'Cc1sc(Cl)cc1C1=C(c2cc(Cl)sc2C)C(F)(F)C(F)(F)C1(F)F': {0: 5,\n",
       "              1: 0},\n",
       "             'Cc1sc(-c2ccc(CO)cc2)cc1C1=C(c2cc(-c3ccc(CO)cc3)sc2C)C(F)(F)C(F)(F)C1(F)F': {0: 1,\n",
       "              1: 0},\n",
       "             'CC1(C)c2c(Br)ccc(Br)c2C(C)(C)N1[O]': {0: 2, 1: 0},\n",
       "             'CC1(C)c2c(-c3ccncc3)ccc(-c3ccncc3)c2C(C)(C)N1[O]': {0: 1, 1: 0},\n",
       "             'CCOC(=O)c1nc2ccccc2c(Br)c1C': {0: 1, 1: 0},\n",
       "             'CCOC(=O)c1nc2ccccc2c(-c2ccccc2)c1C': {0: 1, 1: 0},\n",
       "             'C=C(Br)C(=O)OCC': {0: 9, 1: 0},\n",
       "             'C=C(C(=O)OCC)c1ccccc1C': {0: 1, 1: 0},\n",
       "             'Cc1cccc(C)c1Br': {0: 38, 1: 2},\n",
       "             'Cc1cccc(C)c1-c1ccccc1': {0: 8, 1: 1},\n",
       "             'Brc1ccnc(-c2cc(Br)ccn2)c1': {0: 18, 1: 1},\n",
       "             'CCCCCCn1c2ccccc2c2cc(B3OC(C)(C)C(C)(C)O3)ccc21': {0: 5, 1: 0},\n",
       "             'CCCCCCn1c2ccccc2c2cc(-c3ccnc(-c4cc(-c5ccc6c(c5)c5ccccc5n6CCCCCC)ccn4)c3)ccc21': {0: 1,\n",
       "              1: 0},\n",
       "             'CC(C)Oc1ccc(Oc2ncc(Br)s2)cc1': {0: 1, 1: 0},\n",
       "             'C[C@@H](C=CB1OC(C)(C)C(C)(C)O1)N1C(=O)c2ccccc2C1=O': {0: 3,\n",
       "              1: 0},\n",
       "             'CC(C)Oc1ccc(Oc2ncc(/C=C/[C@H](C)N3C(=O)c4ccccc4C3=O)s2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'Brc1ccccc1Br': {0: 42, 1: 1},\n",
       "             'c1ccc(-c2ccccc2-c2ccccc2)cc1': {0: 9, 1: 0},\n",
       "             'O=C(C(=O)c1ccc(Br)cc1)c1ccc(Br)cc1': {0: 9, 1: 0},\n",
       "             'CCCC[B-](F)(F)F': {0: 8, 1: 0},\n",
       "             'CCCCc1ccc(C(=O)C(=O)c2ccc(CCCC)cc2)cc1': {0: 1, 1: 0},\n",
       "             'O=[N+]([O-])c1ccc(Br)c(F)c1': {0: 1, 1: 0},\n",
       "             'COc1ccc(B(O)O)c(OC)c1': {0: 57, 1: 4},\n",
       "             'COc1ccc(-c2ccc([N+](=O)[O-])cc2F)c(OC)c1': {0: 1, 1: 0},\n",
       "             'Fc1ccc([B-](c2ccc(F)cc2)(c2ccc(F)cc2)c2ccc(F)cc2)cc1': {0: 5,\n",
       "              1: 1},\n",
       "             'COc1ccc(-c2ccc(F)cc2)cc1': {0: 26, 1: 0},\n",
       "             'Brc1ccccn1': {0: 192, 1: 12},\n",
       "             'OB(O)c1ccc2c(c1)CN1CN2Cc2cc(B(O)O)ccc21': {0: 5, 1: 0},\n",
       "             'c1ccc(-c2ccc3c(c2)CN2CN3Cc3cc(-c4ccccn4)ccc32)nc1': {0: 1, 1: 0},\n",
       "             'CN1OC[C@@H]2CN[C@H](c3ccc(Br)cc3)C[C@@H]21': {0: 1, 1: 0},\n",
       "             'CN1OC[C@H]2CN[C@@H](c3ccc(-c4ccccc4)cc3)C[C@H]21': {0: 1, 1: 0},\n",
       "             'COc1ccc(OC)c(Br)c1': {0: 6, 1: 0},\n",
       "             'COc1ccc(OC)c(-c2ccccc2C=O)c1': {0: 2, 1: 0},\n",
       "             'Brc1ccncc1': {0: 53, 1: 3},\n",
       "             'O=Cc1ccccc1-c1ccncc1': {0: 2, 1: 0},\n",
       "             'Brc1ccc(Br)c2snnc12': {0: 1, 1: 1},\n",
       "             'Cn1c2ccccc2c2cc(B3OC(C)(C)C(C)(C)O3)ccc21': {0: 3, 1: 0},\n",
       "             'Cn1c2ccccc2c2cc(-c3ccc(-c4ccc5c(c4)c4ccccc4n5C)c4nsnc34)ccc21': {0: 1,\n",
       "              1: 0},\n",
       "             'Sc1c(Br)cc(Br)cc1Br': {0: 1, 1: 0},\n",
       "             'Nc1ccc(B(O)O)cc1': {0: 52, 1: 6},\n",
       "             'Nc1ccc(-c2cc(-c3ccc(N)cc3)c(S)c(-c3ccc(N)cc3)c2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'C[Si](C)(C)c1c(Br)cccc1Br': {0: 4, 1: 0},\n",
       "             'CCCCCCNc1ccccc1B1OC(C)(C)C(C)(C)O1': {0: 4, 1: 0},\n",
       "             'CCCCCCNc1ccccc1-c1cccc(-c2ccccc2NCCCCCC)c1[Si](C)(C)C': {0: 1,\n",
       "              1: 0},\n",
       "             'Brc1cnc2c(ccc3cc(Br)cnc32)c1': {0: 6, 1: 1},\n",
       "             'COc1ccc(-c2cnc3c(ccc4cc(-c5ccc(OC)cc5)cnc43)c2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'C=CCOC1(C)CCN(c2c([C@H](OC(C)(C)C)C(=O)OCC)c(C)nc3cc(-c4cccc(Br)c4)nn23)CC1': {0: 2,\n",
       "              1: 0},\n",
       "             'COC(=O)c1ccc(O)c(B2OC(C)(C)C(C)(C)O2)c1': {0: 1, 1: 0},\n",
       "             'C=CCOC1(C)CCN(c2c([C@H](OC(C)(C)C)C(=O)OCC)c(C)nc3cc(-c4cccc(-c5cc(C(=O)OC)ccc5O)c4)nn23)CC1': {0: 1,\n",
       "              1: 0},\n",
       "             'Cc1cccc(Cl)c1S(=O)(=O)Nc1cc(Br)cnc1Cl': {0: 0, 1: 1},\n",
       "             'CNc1nc2ccc(B3OC(C)(C)C(C)(C)O3)cc2s1': {0: 0, 1: 2},\n",
       "             'CNc1nc2ccc(-c3cnc(Cl)c(NS(=O)(=O)c4c(C)cccc4Cl)c3)cc2s1': {0: 0,\n",
       "              1: 1},\n",
       "             'CCOc1c(F)c(F)c([B-](F)(F)F)c(F)c1F': {0: 3, 1: 0},\n",
       "             'CCOc1c(F)c(F)c(-c2ccc(C)cc2)c(F)c1F': {0: 1, 1: 0},\n",
       "             'CC(C)(C)c1ccc(Br)cc1': {0: 28, 1: 0},\n",
       "             'COc1ccc(-c2ccc(C(C)(C)C)cc2)cc1': {0: 6, 1: 1},\n",
       "             'Brc1nnc(SCc2ccccc2)n1Cc1ccccc1': {0: 4, 1: 0},\n",
       "             'c1ccc(CSc2nnc(-c3ccccc3)n2Cc2ccccc2)cc1': {0: 1, 1: 0},\n",
       "             'CC(=O)Oc1ccc([C@H](OC(C)=O)[C@H]2[C@@H](/C=C\\\\I)C(C)=C[C@@H]3[C@@H]2C(C)(C)C[C@@]3(C)O)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'CCOC(=O)C=CC(C)=CC[C@@H](O[Si](CC)(CC)CC)[C@@H](C)[C@@H](C=CC=CB1OC(C)(C)C(C)(C)O1)O[Si](CC)(CC)CC': {0: 1,\n",
       "              1: 0},\n",
       "             'CCOC(=O)C=CC(C)=CC[C@@H](O[Si](CC)(CC)CC)[C@@H](C)[C@@H](/C=C\\\\C=C\\\\C=C/[C@H]1C(C)=C[C@@H]2[C@@H]([C@H]1[C@@H](OC(C)=O)c1ccc(OC(C)=O)cc1)C(C)(C)C[C@@]2(C)O)O[Si](CC)(CC)CC': {0: 1,\n",
       "              1: 0},\n",
       "             'CC(C)(C)[Si](C)(C)Oc1ccc(Nc2ccc(Br)cn2)cc1': {0: 1, 1: 0},\n",
       "             'CC(C)(C)[Si](C)(C)Oc1ccc(B(O)O)cc1': {0: 35, 1: 0},\n",
       "             'CC(C)(C)[Si](C)(C)Oc1ccc(Nc2ccc(-c3ccc(O[Si](C)(C)C(C)(C)C)cc3)cn2)cc1': {0: 1,\n",
       "              1: 0},\n",
       "             'Ic1nccn1-c1ccccc1-n1ccnc1I': {0: 2, 1: 0},\n",
       "             'c1ccc(-c2ncc3c4nccn4c4ccccc4n23)cc1': {0: 1, 1: 0},\n",
       "             'O=Cc1cc(Br)c(-c2cc(Br)c(Br)s2)s1': {0: 2, 1: 0},\n",
       "             'O=Cc1cc(-c2ccc(N(c3ccccc3)c3ccccc3)cc2)c(-c2cc(-c3ccc(N(c4ccccc4)c4ccccc4)cc3)c(-c3ccc(N(c4ccccc4)c4ccccc4)cc3)s2)s1': {0: 1,\n",
       "              1: 0},\n",
       "             'COc1ccccc1Br': {0: 54, 1: 6},\n",
       "             'C=C/C=C/B1OC(C)CC(C)(C)O1': {0: 3, 1: 4},\n",
       "             'C=C/C=C/c1ccccc1OC': {0: 0, 1: 2},\n",
       "             'Clc1ncccn1': {0: 44, 1: 2},\n",
       "             'Cc1ccccc1-c1ncccn1': {0: 2, 1: 0},\n",
       "             'Nc1ccc(-c2ccccc2)cc1Br': {0: 1, 1: 0},\n",
       "             'Nc1ccc(-c2ccccc2)cc1-c1ccccc1': {0: 2, 1: 0},\n",
       "             'N#CCC/C=C/[B-](F)(F)F': {0: 1, 1: 0},\n",
       "             'C=C(/C=C/CCC#N)CCO': {0: 1, 1: 0},\n",
       "             'CCOC(=O)c1c(CC)c2cc3nc(c(OC)c4[nH]c(cc5nc(c(Br)c1[nH]2)CC5(C)C)c(CC)c4C(=O)OCC)CC3(C)C': {0: 4,\n",
       "              1: 0},\n",
       "             'CC(C)(C)OC(=O)CCc1ccc(B2OC(C)(C)C(C)(C)O2)cc1': {0: 2, 1: 0},\n",
       "             'CCOC(=O)c1c(CC)c2cc3nc(c(-c4cccc(CCC(=O)OC(C)(C)C)c4)c4[nH]c(cc5nc(c(OC)c1[nH]2)CC5(C)C)c(CC)c4C(=O)OCC)CC3(C)C': {0: 1,\n",
       "              1: 0},\n",
       "             'CCCCC#CC(OC(=O)OC)c1ccccc1': {0: 11, 1: 0},\n",
       "             'CB(O)O': {0: 84, 1: 17},\n",
       "             'CCCCC(C)=C=Cc1ccccc1': {0: 1, 1: 0},\n",
       "             'C[C@@H]1ON(C)[C@H]2C[C@H]3c4cc(Br)ccc4CCN3C(=O)[C@@H]12': {0: 3,\n",
       "              1: 0},\n",
       "             'C[C@H]1ON(C)[C@@H]2C[C@@H]3c4cc(-c5ccccc5)ccc4CCN3C(=O)[C@H]12': {0: 1,\n",
       "              1: 0},\n",
       "             'O=[N+]([O-])c1ccc2c(I)nn(C3CCCCO3)c2c1': {0: 11, 1: 0},\n",
       "             'O=[N+]([O-])c1ccc2c(-c3cccnc3)nn(C3CCCCO3)c2c1': {0: 1, 1: 0},\n",
       "             'CC(C)[C@H]1NC(=O)C(I)=CN1C(=O)c1ccccc1': {0: 13, 1: 3},\n",
       "             'Oc1cccc([B-](F)(F)F)c1': {0: 5, 1: 1},\n",
       "             'CC(C)[C@H]1NC(=O)C(c2cccc(O)c2)=CN1C(=O)c1ccccc1': {0: 1, 1: 0},\n",
       "             'OC[C@H]1[C@H](I)[C@@H]1CCc1ccccc1': {0: 8, 1: 0},\n",
       "             'Cc1cc(C)c(B(O)O)c(C)c1': {0: 96, 1: 9},\n",
       "             'Cc1cc(C)c([C@@H]2[C@H](CCc3ccccc3)[C@H]2CO)c(C)c1': {0: 1, 1: 0},\n",
       "             'COc1ccc2cc(Br)ccc2c1': {0: 17, 1: 0},\n",
       "             'COc1ccc2cc(B3OC(C)(C)C(C)(C)O3)ccc2c1': {0: 1, 1: 0},\n",
       "             'Clc1cc(NC2CCOC2)ncn1': {0: 1, 1: 0},\n",
       "             ...})"
      ]
     },
     "execution_count": 152,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "class_mols"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 155,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Now calculate the mean class for each molecule\n",
    "positive_fractions = {mol: class_counts[1] / (class_counts[0] + class_counts[1]) for mol, class_counts in class_mols.items() if (class_counts[0] + class_counts[1]) > 10}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 156,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 864x864 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "plt.figure(figsize=(12,12))\n",
    "sns.histplot(positive_fractions.values(), bins=20)\n",
    "plt.title(\"Yield class concentrations of molecules, $\\\\frac{N_1}{N_0 + N_1}$\\n values close to 1 means the molecules is mainly in yieldclass 1,\\n and values close to 0 that it's mainly in yield class 0\", fontsize=20);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see that our dataset has a pronounced peak at 0, and a less pronounced peak at 1. This means that the vast majority of molecules occur only in one of the yield classes. This means that any statistical model will likely use this information when making its prediction, simply by looking at the identity of the molecules in the reaction."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Predicting Yields without any chemical information\n",
    "\n",
    "To test whether this works, we can create a new dataset where the feature of each molecule is simply a unique symbol. We can then use decision trees with support for categorical variables to test how well it can learn the yield classes from only a symbolic value for a molecule. You can think of replacing each molecule by a distinct nonsense-word which has no correlation to the molecules chemical properties. The only thing driving the statistical model will then be how often the molecule occurs in positive vs. negative reactinos"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Replace each molecule with its unique integer index\n",
    "symbolic_reactions = data[['Component_1', 'Component_2', 'Product']].applymap(lambda x: mol_id[canonicalize_smiles(x)]).values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[69301, 72252, 74192],\n",
       "       [57022, 51594, 51017],\n",
       "       [46375, 72030, 46307],\n",
       "       ...,\n",
       "       [31483, 40564, 31474],\n",
       "       [ 2995, 46667,  2687],\n",
       "       [ 9548, 32221, 32009]], dtype=int64)"
      ]
     },
     "execution_count": 87,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "symbolic_reactions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 150,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "40ad8ac9453e4f53b2217c1740946a67",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "MetricVisualizer(layout=Layout(align_self='stretch', height='500px'))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from catboost import CatBoostClassifier, Pool, metrics, cv\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "#X_train, X_validation, y_train, y_validation = train_test_split(properties.values, y, train_size=0.75, random_state=42)\n",
    "\n",
    "model = CatBoostClassifier(\n",
    "    #custom_loss=[metrics.Accuracy()],\n",
    "    custom_metric=[metrics.BalancedAccuracy(), metrics.AUC(), metrics.Precision(), metrics.Recall()],\n",
    "    random_seed=42,\n",
    "    logging_level='Silent'\n",
    ")\n",
    "cv_params = model.get_params()\n",
    "cv_params.update({\n",
    "    'loss_function': metrics.Logloss(),\n",
    "    'iterations': 2000\n",
    "})\n",
    "cv_data = cv(\n",
    "    Pool(properties.values, y),\n",
    "    cv_params,\n",
    "    plot=True,\n",
    "    stratified=True,\n",
    "    fold_count=3\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 157,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0dcc683066d24d5eb6454365634b3b18",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "MetricVisualizer(layout=Layout(align_self='stretch', height='500px'))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from catboost import CatBoostClassifier, Pool, metrics, cv\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "X_train, X_validation, y_train, y_validation = train_test_split(symbolic_reactions, y, train_size=0.75, random_state=42)\n",
    "\n",
    "catergorical_feature_indices = [0,1,2]\n",
    "model = CatBoostClassifier(\n",
    "    #custom_loss=[metrics.Accuracy()],\n",
    "    custom_metric=[metrics.Accuracy(), metrics.BalancedAccuracy(), metrics.AUC(), metrics.Precision(), metrics.Recall()],\n",
    "    random_seed=42,\n",
    "    logging_level='Silent'\n",
    ")\n",
    "cv_params = model.get_params()\n",
    "cv_params.update({\n",
    "    'loss_function': metrics.Logloss()\n",
    "})\n",
    "cv_data = cv(\n",
    "    Pool(symbolic_reactions, y, cat_features=catergorical_feature_indices),\n",
    "    cv_params,\n",
    "    plot=True,\n",
    "    stratified=True,\n",
    "    fold_count=3\n",
    ")\n",
    "#model.fit(X_train, y_train, cat_features=catergorical_feature_indices, eval_set=(X_validation, y_validation), plot=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Predicting yields from Random Vectors\n",
    "\n",
    "A consequence of this is that as long as the features describing molecules are distinct enough, any statistical model will pick up on these. So even if we provide the model with chemically meaningful descriptors, the underlying sampling bias will drive much of the learning. It's not necessarily  the case that the actual descriptors are more useful than as unique identifiers. We can test this by essentially creating random vectors for each molecule. If you are familiar with NLP, you should see that this is very close to representing each molecule by an embedding vector."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "75874"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "n_mols = max(mol_id.values())+1\n",
    "n_mols"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "embedding_dim = 200\n",
    "embedding_matrix = np.random.randn(n_mols, embedding_dim)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "embedded_reactions = embedding_matrix[symbolic_reactions].reshape(-1, 3*embedding_dim)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(54297, 600)"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "embedded_reactions.reshape(-1, 3*embedding_dim).shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "embedded_reactions = embedding_matrix[symbolic_reactions].sum(axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Erik\\AppData\\Local\\Temp\\ipykernel_24528\\3186444372.py:1: FutureWarning: Dropping of nuisance columns in DataFrame reductions (with 'numeric_only=None') is deprecated; in a future version this will raise TypeError.  Select only valid columns before calling the reduction.\n",
      "  data.loc[:, data.std() > 1e2].hist(bins=100, figsize=(16,8))\n"
     ]
    },
    {
     "ename": "IndexingError",
     "evalue": "Unalignable boolean Series provided as indexer (index of the boolean Series and of the indexed object do not match).",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m\n",
      "\u001b[1;31mIndexingError\u001b[0m                             Traceback (most recent call last)\n",
      "\u001b[1;32mf:\\programming_projects\\yield_prediction\\yield_prediction.ipynb Cell 24'\u001b[0m in \u001b[0;36m<cell line: 1>\u001b[1;34m()\u001b[0m\n",
      "\u001b[1;32m----> <a href='vscode-notebook-cell:/f%3A/programming_projects/yield_prediction/yield_prediction.ipynb#ch0000023?line=0'>1</a>\u001b[0m data\u001b[39m.\u001b[39;49mloc[:, data\u001b[39m.\u001b[39;49mstd() \u001b[39m>\u001b[39;49m \u001b[39m1e2\u001b[39;49m]\u001b[39m.\u001b[39mhist(bins\u001b[39m=\u001b[39m\u001b[39m100\u001b[39m, figsize\u001b[39m=\u001b[39m(\u001b[39m16\u001b[39m,\u001b[39m8\u001b[39m))\n",
      "\n",
      "File \u001b[1;32mf:\\Anaconda\\envs\\yield-prediction\\lib\\site-packages\\pandas\\core\\indexing.py:961\u001b[0m, in \u001b[0;36m_LocationIndexer.__getitem__\u001b[1;34m(self, key)\u001b[0m\n",
      "\u001b[0;32m    959\u001b[0m     \u001b[39mif\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_is_scalar_access(key):\n",
      "\u001b[0;32m    960\u001b[0m         \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39mobj\u001b[39m.\u001b[39m_get_value(\u001b[39m*\u001b[39mkey, takeable\u001b[39m=\u001b[39m\u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_takeable)\n",
      "\u001b[1;32m--> 961\u001b[0m     \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39;49m\u001b[39m.\u001b[39;49m_getitem_tuple(key)\n",
      "\u001b[0;32m    962\u001b[0m \u001b[39melse\u001b[39;00m:\n",
      "\u001b[0;32m    963\u001b[0m     \u001b[39m# we by definition only have the 0th axis\u001b[39;00m\n",
      "\u001b[0;32m    964\u001b[0m     axis \u001b[39m=\u001b[39m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39maxis \u001b[39mor\u001b[39;00m \u001b[39m0\u001b[39m\n",
      "\n",
      "File \u001b[1;32mf:\\Anaconda\\envs\\yield-prediction\\lib\\site-packages\\pandas\\core\\indexing.py:1149\u001b[0m, in \u001b[0;36m_LocIndexer._getitem_tuple\u001b[1;34m(self, tup)\u001b[0m\n",
      "\u001b[0;32m   1146\u001b[0m \u001b[39mif\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_multi_take_opportunity(tup):\n",
      "\u001b[0;32m   1147\u001b[0m     \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_multi_take(tup)\n",
      "\u001b[1;32m-> 1149\u001b[0m \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39;49m\u001b[39m.\u001b[39;49m_getitem_tuple_same_dim(tup)\n",
      "\n",
      "File \u001b[1;32mf:\\Anaconda\\envs\\yield-prediction\\lib\\site-packages\\pandas\\core\\indexing.py:827\u001b[0m, in \u001b[0;36m_LocationIndexer._getitem_tuple_same_dim\u001b[1;34m(self, tup)\u001b[0m\n",
      "\u001b[0;32m    824\u001b[0m \u001b[39mif\u001b[39;00m com\u001b[39m.\u001b[39mis_null_slice(key):\n",
      "\u001b[0;32m    825\u001b[0m     \u001b[39mcontinue\u001b[39;00m\n",
      "\u001b[1;32m--> 827\u001b[0m retval \u001b[39m=\u001b[39m \u001b[39mgetattr\u001b[39;49m(retval, \u001b[39mself\u001b[39;49m\u001b[39m.\u001b[39;49mname)\u001b[39m.\u001b[39;49m_getitem_axis(key, axis\u001b[39m=\u001b[39;49mi)\n",
      "\u001b[0;32m    828\u001b[0m \u001b[39m# We should never have retval.ndim < self.ndim, as that should\u001b[39;00m\n",
      "\u001b[0;32m    829\u001b[0m \u001b[39m#  be handled by the _getitem_lowerdim call above.\u001b[39;00m\n",
      "\u001b[0;32m    830\u001b[0m \u001b[39massert\u001b[39;00m retval\u001b[39m.\u001b[39mndim \u001b[39m==\u001b[39m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39mndim\n",
      "\n",
      "File \u001b[1;32mf:\\Anaconda\\envs\\yield-prediction\\lib\\site-packages\\pandas\\core\\indexing.py:1182\u001b[0m, in \u001b[0;36m_LocIndexer._getitem_axis\u001b[1;34m(self, key, axis)\u001b[0m\n",
      "\u001b[0;32m   1180\u001b[0m     \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_get_slice_axis(key, axis\u001b[39m=\u001b[39maxis)\n",
      "\u001b[0;32m   1181\u001b[0m \u001b[39melif\u001b[39;00m com\u001b[39m.\u001b[39mis_bool_indexer(key):\n",
      "\u001b[1;32m-> 1182\u001b[0m     \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39;49m\u001b[39m.\u001b[39;49m_getbool_axis(key, axis\u001b[39m=\u001b[39;49maxis)\n",
      "\u001b[0;32m   1183\u001b[0m \u001b[39melif\u001b[39;00m is_list_like_indexer(key):\n",
      "\u001b[0;32m   1184\u001b[0m \n",
      "\u001b[0;32m   1185\u001b[0m     \u001b[39m# an iterable multi-selection\u001b[39;00m\n",
      "\u001b[0;32m   1186\u001b[0m     \u001b[39mif\u001b[39;00m \u001b[39mnot\u001b[39;00m (\u001b[39misinstance\u001b[39m(key, \u001b[39mtuple\u001b[39m) \u001b[39mand\u001b[39;00m \u001b[39misinstance\u001b[39m(labels, MultiIndex)):\n",
      "\n",
      "File \u001b[1;32mf:\\Anaconda\\envs\\yield-prediction\\lib\\site-packages\\pandas\\core\\indexing.py:984\u001b[0m, in \u001b[0;36m_LocationIndexer._getbool_axis\u001b[1;34m(self, key, axis)\u001b[0m\n",
      "\u001b[0;32m    981\u001b[0m \u001b[39mdef\u001b[39;00m \u001b[39m_getbool_axis\u001b[39m(\u001b[39mself\u001b[39m, key, axis: \u001b[39mint\u001b[39m):\n",
      "\u001b[0;32m    982\u001b[0m     \u001b[39m# caller is responsible for ensuring non-None axis\u001b[39;00m\n",
      "\u001b[0;32m    983\u001b[0m     labels \u001b[39m=\u001b[39m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39mobj\u001b[39m.\u001b[39m_get_axis(axis)\n",
      "\u001b[1;32m--> 984\u001b[0m     key \u001b[39m=\u001b[39m check_bool_indexer(labels, key)\n",
      "\u001b[0;32m    985\u001b[0m     inds \u001b[39m=\u001b[39m key\u001b[39m.\u001b[39mnonzero()[\u001b[39m0\u001b[39m]\n",
      "\u001b[0;32m    986\u001b[0m     \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39mobj\u001b[39m.\u001b[39m_take_with_is_copy(inds, axis\u001b[39m=\u001b[39maxis)\n",
      "\n",
      "File \u001b[1;32mf:\\Anaconda\\envs\\yield-prediction\\lib\\site-packages\\pandas\\core\\indexing.py:2383\u001b[0m, in \u001b[0;36mcheck_bool_indexer\u001b[1;34m(index, key)\u001b[0m\n",
      "\u001b[0;32m   2381\u001b[0m     mask \u001b[39m=\u001b[39m isna(result\u001b[39m.\u001b[39m_values)\n",
      "\u001b[0;32m   2382\u001b[0m     \u001b[39mif\u001b[39;00m mask\u001b[39m.\u001b[39many():\n",
      "\u001b[1;32m-> 2383\u001b[0m         \u001b[39mraise\u001b[39;00m IndexingError(\n",
      "\u001b[0;32m   2384\u001b[0m             \u001b[39m\"\u001b[39m\u001b[39mUnalignable boolean Series provided as \u001b[39m\u001b[39m\"\u001b[39m\n",
      "\u001b[0;32m   2385\u001b[0m             \u001b[39m\"\u001b[39m\u001b[39mindexer (index of the boolean Series and of \u001b[39m\u001b[39m\"\u001b[39m\n",
      "\u001b[0;32m   2386\u001b[0m             \u001b[39m\"\u001b[39m\u001b[39mthe indexed object do not match).\u001b[39m\u001b[39m\"\u001b[39m\n",
      "\u001b[0;32m   2387\u001b[0m         )\n",
      "\u001b[0;32m   2388\u001b[0m     \u001b[39mreturn\u001b[39;00m result\u001b[39m.\u001b[39mastype(\u001b[39mbool\u001b[39m)\u001b[39m.\u001b[39m_values\n",
      "\u001b[0;32m   2389\u001b[0m \u001b[39mif\u001b[39;00m is_object_dtype(key):\n",
      "\u001b[0;32m   2390\u001b[0m     \u001b[39m# key might be object-dtype bool, check_array_indexer needs bool array\u001b[39;00m\n",
      "\n",
      "\u001b[1;31mIndexingError\u001b[0m: Unalignable boolean Series provided as indexer (index of the boolean Series and of the indexed object do not match)."
     ]
    }
   ],
   "source": [
    "data.loc[:, data.std() > 1e2].hist(bins=100, figsize=(16,8))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "ename": "KeyError",
     "evalue": "'Ipc_(C1_RDkit)'",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m\n",
      "\u001b[1;31mKeyError\u001b[0m                                  Traceback (most recent call last)\n",
      "File \u001b[1;32mf:\\Anaconda\\envs\\yield-prediction\\lib\\site-packages\\pandas\\core\\indexes\\base.py:3621\u001b[0m, in \u001b[0;36mIndex.get_loc\u001b[1;34m(self, key, method, tolerance)\u001b[0m\n",
      "\u001b[0;32m   <a href='file:///f%3A/Anaconda/envs/yield-prediction/lib/site-packages/pandas/core/indexes/base.py?line=3619'>3620</a>\u001b[0m \u001b[39mtry\u001b[39;00m:\n",
      "\u001b[1;32m-> <a href='file:///f%3A/Anaconda/envs/yield-prediction/lib/site-packages/pandas/core/indexes/base.py?line=3620'>3621</a>\u001b[0m     \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39;49m\u001b[39m.\u001b[39;49m_engine\u001b[39m.\u001b[39;49mget_loc(casted_key)\n",
      "\u001b[0;32m   <a href='file:///f%3A/Anaconda/envs/yield-prediction/lib/site-packages/pandas/core/indexes/base.py?line=3621'>3622</a>\u001b[0m \u001b[39mexcept\u001b[39;00m \u001b[39mKeyError\u001b[39;00m \u001b[39mas\u001b[39;00m err:\n",
      "\n",
      "File \u001b[1;32mf:\\Anaconda\\envs\\yield-prediction\\lib\\site-packages\\pandas\\_libs\\index.pyx:136\u001b[0m, in \u001b[0;36mpandas._libs.index.IndexEngine.get_loc\u001b[1;34m()\u001b[0m\n",
      "\n",
      "File \u001b[1;32mf:\\Anaconda\\envs\\yield-prediction\\lib\\site-packages\\pandas\\_libs\\index.pyx:163\u001b[0m, in \u001b[0;36mpandas._libs.index.IndexEngine.get_loc\u001b[1;34m()\u001b[0m\n",
      "\n",
      "File \u001b[1;32mpandas\\_libs\\hashtable_class_helper.pxi:5198\u001b[0m, in \u001b[0;36mpandas._libs.hashtable.PyObjectHashTable.get_item\u001b[1;34m()\u001b[0m\n",
      "\n",
      "File \u001b[1;32mpandas\\_libs\\hashtable_class_helper.pxi:5206\u001b[0m, in \u001b[0;36mpandas._libs.hashtable.PyObjectHashTable.get_item\u001b[1;34m()\u001b[0m\n",
      "\n",
      "\u001b[1;31mKeyError\u001b[0m: 'Ipc_(C1_RDkit)'\n",
      "\n",
      "The above exception was the direct cause of the following exception:\n",
      "\n",
      "\u001b[1;31mKeyError\u001b[0m                                  Traceback (most recent call last)\n",
      "\u001b[1;32mf:\\programming_projects\\yield_prediction\\yield_prediction.ipynb Cell 25'\u001b[0m in \u001b[0;36m<cell line: 1>\u001b[1;34m()\u001b[0m\n",
      "\u001b[1;32m----> <a href='vscode-notebook-cell:/f%3A/programming_projects/yield_prediction/yield_prediction.ipynb#ch0000022?line=0'>1</a>\u001b[0m data[\u001b[39m'\u001b[39;49m\u001b[39mIpc_(C1_RDkit)\u001b[39;49m\u001b[39m'\u001b[39;49m]\u001b[39m.\u001b[39mapply(\u001b[39mlambda\u001b[39;00m x: np\u001b[39m.\u001b[39mlog(x))\u001b[39m.\u001b[39mhist(bins\u001b[39m=\u001b[39m\u001b[39m100\u001b[39m)\n",
      "\n",
      "File \u001b[1;32mf:\\Anaconda\\envs\\yield-prediction\\lib\\site-packages\\pandas\\core\\frame.py:3505\u001b[0m, in \u001b[0;36mDataFrame.__getitem__\u001b[1;34m(self, key)\u001b[0m\n",
      "\u001b[0;32m   <a href='file:///f%3A/Anaconda/envs/yield-prediction/lib/site-packages/pandas/core/frame.py?line=3502'>3503</a>\u001b[0m \u001b[39mif\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39mcolumns\u001b[39m.\u001b[39mnlevels \u001b[39m>\u001b[39m \u001b[39m1\u001b[39m:\n",
      "\u001b[0;32m   <a href='file:///f%3A/Anaconda/envs/yield-prediction/lib/site-packages/pandas/core/frame.py?line=3503'>3504</a>\u001b[0m     \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_getitem_multilevel(key)\n",
      "\u001b[1;32m-> <a href='file:///f%3A/Anaconda/envs/yield-prediction/lib/site-packages/pandas/core/frame.py?line=3504'>3505</a>\u001b[0m indexer \u001b[39m=\u001b[39m \u001b[39mself\u001b[39;49m\u001b[39m.\u001b[39;49mcolumns\u001b[39m.\u001b[39;49mget_loc(key)\n",
      "\u001b[0;32m   <a href='file:///f%3A/Anaconda/envs/yield-prediction/lib/site-packages/pandas/core/frame.py?line=3505'>3506</a>\u001b[0m \u001b[39mif\u001b[39;00m is_integer(indexer):\n",
      "\u001b[0;32m   <a href='file:///f%3A/Anaconda/envs/yield-prediction/lib/site-packages/pandas/core/frame.py?line=3506'>3507</a>\u001b[0m     indexer \u001b[39m=\u001b[39m [indexer]\n",
      "\n",
      "File \u001b[1;32mf:\\Anaconda\\envs\\yield-prediction\\lib\\site-packages\\pandas\\core\\indexes\\base.py:3623\u001b[0m, in \u001b[0;36mIndex.get_loc\u001b[1;34m(self, key, method, tolerance)\u001b[0m\n",
      "\u001b[0;32m   <a href='file:///f%3A/Anaconda/envs/yield-prediction/lib/site-packages/pandas/core/indexes/base.py?line=3620'>3621</a>\u001b[0m     \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_engine\u001b[39m.\u001b[39mget_loc(casted_key)\n",
      "\u001b[0;32m   <a href='file:///f%3A/Anaconda/envs/yield-prediction/lib/site-packages/pandas/core/indexes/base.py?line=3621'>3622</a>\u001b[0m \u001b[39mexcept\u001b[39;00m \u001b[39mKeyError\u001b[39;00m \u001b[39mas\u001b[39;00m err:\n",
      "\u001b[1;32m-> <a href='file:///f%3A/Anaconda/envs/yield-prediction/lib/site-packages/pandas/core/indexes/base.py?line=3622'>3623</a>\u001b[0m     \u001b[39mraise\u001b[39;00m \u001b[39mKeyError\u001b[39;00m(key) \u001b[39mfrom\u001b[39;00m \u001b[39merr\u001b[39;00m\n",
      "\u001b[0;32m   <a href='file:///f%3A/Anaconda/envs/yield-prediction/lib/site-packages/pandas/core/indexes/base.py?line=3623'>3624</a>\u001b[0m \u001b[39mexcept\u001b[39;00m \u001b[39mTypeError\u001b[39;00m:\n",
      "\u001b[0;32m   <a href='file:///f%3A/Anaconda/envs/yield-prediction/lib/site-packages/pandas/core/indexes/base.py?line=3624'>3625</a>\u001b[0m     \u001b[39m# If we have a listlike key, _check_indexing_error will raise\u001b[39;00m\n",
      "\u001b[0;32m   <a href='file:///f%3A/Anaconda/envs/yield-prediction/lib/site-packages/pandas/core/indexes/base.py?line=3625'>3626</a>\u001b[0m     \u001b[39m#  InvalidIndexError. Otherwise we fall through and re-raise\u001b[39;00m\n",
      "\u001b[0;32m   <a href='file:///f%3A/Anaconda/envs/yield-prediction/lib/site-packages/pandas/core/indexes/base.py?line=3626'>3627</a>\u001b[0m     \u001b[39m#  the TypeError.\u001b[39;00m\n",
      "\u001b[0;32m   <a href='file:///f%3A/Anaconda/envs/yield-prediction/lib/site-packages/pandas/core/indexes/base.py?line=3627'>3628</a>\u001b[0m     \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_check_indexing_error(key)\n",
      "\n",
      "\u001b[1;31mKeyError\u001b[0m: 'Ipc_(C1_RDkit)'"
     ]
    }
   ],
   "source": [
    "\n",
    "data['Ipc_(C1_RDkit)'].apply(lambda x: np.log(x)).hist(bins=100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "f:\\Anaconda\\envs\\yield-prediction\\lib\\site-packages\\pandas\\core\\arraylike.py:397: RuntimeWarning: invalid value encountered in log\n",
      "  result = getattr(ufunc, method)(*inputs, **kwargs)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([[<AxesSubplot:title={'center':'Kappa1_(C1_rdkit)'}>,\n",
       "        <AxesSubplot:title={'center':'Kappa2_(C1_rdkit)'}>,\n",
       "        <AxesSubplot:title={'center':'Kappa3_(C1_rdkit)'}>],\n",
       "       [<AxesSubplot:title={'center':'Kappa1_(C2_rdkit)'}>,\n",
       "        <AxesSubplot:title={'center':'Kappa2_(C2_rdkit)'}>,\n",
       "        <AxesSubplot:title={'center':'Kappa3_(C2_rdkit)'}>],\n",
       "       [<AxesSubplot:title={'center':'Kappa1_(P_rdkit)'}>,\n",
       "        <AxesSubplot:title={'center':'Kappa2_(P_rdkit)'}>,\n",
       "        <AxesSubplot:title={'center':'Kappa3_(P_rdkit)'}>]], dtype=object)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1440x720 with 9 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "data[[c for c in data.columns if 'Kappa' in c]].apply(lambda x: np.log(np.log(x))).hist(bins=100, figsize=(20, 10))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['Ipc_(C1_rdkit)',\n",
       " 'Kappa1_(C1_rdkit)',\n",
       " 'Kappa2_(C1_rdkit)',\n",
       " 'Kappa3_(C1_rdkit)',\n",
       " 'Ipc_(C2_rdkit)',\n",
       " 'Kappa1_(C2_rdkit)',\n",
       " 'Kappa2_(C2_rdkit)',\n",
       " 'Kappa3_(C2_rdkit)',\n",
       " 'Ipc_(P_rdkit)',\n",
       " 'Kappa1_(P_rdkit)',\n",
       " 'Kappa2_(P_rdkit)',\n",
       " 'Kappa3_(P_rdkit)']"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "log_log_columns = [c for c in data.columns if 'Kappa' in c or 'Ipc' in c]\n",
    "log_log_columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "f:\\Anaconda\\envs\\yield-prediction\\lib\\site-packages\\pandas\\core\\arraylike.py:397: RuntimeWarning: invalid value encountered in log\n",
      "  result = getattr(ufunc, method)(*inputs, **kwargs)\n"
     ]
    }
   ],
   "source": [
    "data[log_log_columns] = data[log_log_columns].apply(lambda x: np.log(np.log(x)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'X' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m\n",
      "\u001b[1;31mNameError\u001b[0m                                 Traceback (most recent call last)\n",
      "\u001b[1;32mf:\\programming_projects\\yield_prediction\\yield_prediction.ipynb Cell 29'\u001b[0m in \u001b[0;36m<cell line: 1>\u001b[1;34m()\u001b[0m\n",
      "\u001b[1;32m----> <a href='vscode-notebook-cell:/f%3A/programming_projects/yield_prediction/yield_prediction.ipynb#ch0000028?line=0'>1</a>\u001b[0m X \u001b[39m=\u001b[39m X\u001b[39m.\u001b[39mdrop(columns\u001b[39m=\u001b[39mlog_log_columns)\n",
      "\n",
      "\u001b[1;31mNameError\u001b[0m: name 'X' is not defined"
     ]
    }
   ],
   "source": [
    "X = X.drop(columns=log_log_columns)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from collections import defaultdict\n",
    "\n",
    "def do_cv_training(X,\n",
    "                   y,\n",
    "                   classifier,\n",
    "                   num_splits=10,\n",
    "                   metrics=(\"precision\", 'recall', 'recall_weighted', \"f1\", \"accuracy\", \"roc_auc\", \"roc_auc_ovo_weighted\", \"average_precision\", \"balanced_accuracy\")\n",
    "                  ):\n",
    "        collected_metrics = {metric_name: [] for metric_name in metrics}\n",
    "        scores = cross_validate(classifier, X, y, cv=num_splits, scoring=metrics)\n",
    "        for metric_name in metrics:\n",
    "            collected_metrics[metric_name].extend(scores[\"test_\" + metric_name])\n",
    "        return collected_metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "classifier = RandomForestClassifier(n_estimators=300,class_weight='balanced', n_jobs=14)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "classifier = GradientBoostingClassifier()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "classifier = LogisticRegression()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "y = data['yieldclass'].values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "component_1_columns = symbolic_reactions[:,0]\n",
    "component_2_columns = symbolic_reactions[:,1] + n_mols-1\n",
    "product_columns = symbolic_reactions[:,2] + 2*(n_mols-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "from scipy.sparse import lil_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([69301, 57022, 46375, ..., 31483,  2995,  9548], dtype=int64)"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "symbolic_reactions[:,0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "row_indices = np.arange(len(y))\n",
    "reaction_matrix = lil_matrix((len(y), n_mols*3+3))\n",
    "reaction_matrix[row_indices, component_1_columns] = 1\n",
    "reaction_matrix[row_indices, component_2_columns] = 1\n",
    "reaction_matrix[row_indices, product_columns] = 1\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>iterations</th>\n",
       "      <th>test-Logloss-mean</th>\n",
       "      <th>test-Logloss-std</th>\n",
       "      <th>train-Logloss-mean</th>\n",
       "      <th>train-Logloss-std</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0.665157</td>\n",
       "      <td>0.000061</td>\n",
       "      <td>0.665838</td>\n",
       "      <td>0.000032</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>0.639363</td>\n",
       "      <td>0.000312</td>\n",
       "      <td>0.640955</td>\n",
       "      <td>0.000178</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>0.614341</td>\n",
       "      <td>0.000203</td>\n",
       "      <td>0.616962</td>\n",
       "      <td>0.000142</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>0.591377</td>\n",
       "      <td>0.000230</td>\n",
       "      <td>0.594906</td>\n",
       "      <td>0.000207</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>0.570964</td>\n",
       "      <td>0.000329</td>\n",
       "      <td>0.575022</td>\n",
       "      <td>0.000163</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>995</th>\n",
       "      <td>995</td>\n",
       "      <td>0.240464</td>\n",
       "      <td>0.000812</td>\n",
       "      <td>0.258036</td>\n",
       "      <td>0.001190</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>996</th>\n",
       "      <td>996</td>\n",
       "      <td>0.240465</td>\n",
       "      <td>0.000814</td>\n",
       "      <td>0.258031</td>\n",
       "      <td>0.001191</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>997</th>\n",
       "      <td>997</td>\n",
       "      <td>0.240468</td>\n",
       "      <td>0.000813</td>\n",
       "      <td>0.258026</td>\n",
       "      <td>0.001192</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>998</th>\n",
       "      <td>998</td>\n",
       "      <td>0.240467</td>\n",
       "      <td>0.000815</td>\n",
       "      <td>0.258019</td>\n",
       "      <td>0.001197</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>999</th>\n",
       "      <td>999</td>\n",
       "      <td>0.240472</td>\n",
       "      <td>0.000815</td>\n",
       "      <td>0.258014</td>\n",
       "      <td>0.001198</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1000 rows × 5 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     iterations  test-Logloss-mean  test-Logloss-std  train-Logloss-mean  \\\n",
       "0             0           0.665157          0.000061            0.665838   \n",
       "1             1           0.639363          0.000312            0.640955   \n",
       "2             2           0.614341          0.000203            0.616962   \n",
       "3             3           0.591377          0.000230            0.594906   \n",
       "4             4           0.570964          0.000329            0.575022   \n",
       "..          ...                ...               ...                 ...   \n",
       "995         995           0.240464          0.000812            0.258036   \n",
       "996         996           0.240465          0.000814            0.258031   \n",
       "997         997           0.240468          0.000813            0.258026   \n",
       "998         998           0.240467          0.000815            0.258019   \n",
       "999         999           0.240472          0.000815            0.258014   \n",
       "\n",
       "     train-Logloss-std  \n",
       "0             0.000032  \n",
       "1             0.000178  \n",
       "2             0.000142  \n",
       "3             0.000207  \n",
       "4             0.000163  \n",
       "..                 ...  \n",
       "995           0.001190  \n",
       "996           0.001191  \n",
       "997           0.001192  \n",
       "998           0.001197  \n",
       "999           0.001198  \n",
       "\n",
       "[1000 rows x 5 columns]"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cv_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'precision': [0.6792452830188679,\n",
       "  0.75,\n",
       "  0.7727272727272727,\n",
       "  0.7307692307692307,\n",
       "  0.6491228070175439],\n",
       " 'recall': [0.029008863819500404,\n",
       "  0.0265914585012087,\n",
       "  0.0273972602739726,\n",
       "  0.030620467365028204,\n",
       "  0.029814665592264304],\n",
       " 'recall_weighted': [0.8874769797421731,\n",
       "  0.8877532228360958,\n",
       "  0.8879270651072843,\n",
       "  0.8879270651072843,\n",
       "  0.8872824385302513],\n",
       " 'f1': [0.055641421947449775,\n",
       "  0.051361867704280154,\n",
       "  0.05291828793774319,\n",
       "  0.05877803557617942,\n",
       "  0.05701078582434514],\n",
       " 'accuracy': [0.8874769797421731,\n",
       "  0.8877532228360958,\n",
       "  0.8879270651072843,\n",
       "  0.8879270651072843,\n",
       "  0.8872824385302513],\n",
       " 'roc_auc': [0.763435188498053,\n",
       "  0.7615294199743508,\n",
       "  0.7618612378851163,\n",
       "  0.7528417121469633,\n",
       "  0.7472057914509945],\n",
       " 'roc_auc_ovo_weighted': [0.763435188498053,\n",
       "  0.7615294199743508,\n",
       "  0.7618612378851163,\n",
       "  0.7528417121469633,\n",
       "  0.7472057914509945],\n",
       " 'average_precision': [0.3438266020541892,\n",
       "  0.33367332583274345,\n",
       "  0.3523437463246874,\n",
       "  0.3267970257454841,\n",
       "  0.3154706851902176],\n",
       " 'balanced_accuracy': [0.5136207641688207,\n",
       "  0.5127239442417677,\n",
       "  0.5131787715385251,\n",
       "  0.5145824316446683,\n",
       "  0.5138676155992097]}"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "do_cv_training(embedded_reactions, y, classifier, num_splits=5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'cv_data' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[1;32mf:\\programming_projects\\yield_prediction\\yield_prediction.ipynb Cell 21'\u001b[0m in \u001b[0;36m<cell line: 1>\u001b[1;34m()\u001b[0m\n\u001b[0;32m      <a href='vscode-notebook-cell:/f%3A/programming_projects/yield_prediction/yield_prediction.ipynb#ch0000020?line=0'>1</a>\u001b[0m \u001b[39mprint\u001b[39m(\u001b[39m'\u001b[39m\u001b[39mBest validation accuracy score: \u001b[39m\u001b[39m{:.2f}\u001b[39;00m\u001b[39m±\u001b[39m\u001b[39m{:.2f}\u001b[39;00m\u001b[39m on step \u001b[39m\u001b[39m{}\u001b[39;00m\u001b[39m'\u001b[39m\u001b[39m.\u001b[39mformat(\n\u001b[1;32m----> <a href='vscode-notebook-cell:/f%3A/programming_projects/yield_prediction/yield_prediction.ipynb#ch0000020?line=1'>2</a>\u001b[0m     np\u001b[39m.\u001b[39mmax(cv_data[\u001b[39m'\u001b[39m\u001b[39mtest-Accuracy-mean\u001b[39m\u001b[39m'\u001b[39m]),\n\u001b[0;32m      <a href='vscode-notebook-cell:/f%3A/programming_projects/yield_prediction/yield_prediction.ipynb#ch0000020?line=2'>3</a>\u001b[0m     cv_data[\u001b[39m'\u001b[39m\u001b[39mtest-Accuracy-std\u001b[39m\u001b[39m'\u001b[39m][np\u001b[39m.\u001b[39margmax(cv_data[\u001b[39m'\u001b[39m\u001b[39mtest-Accuracy-mean\u001b[39m\u001b[39m'\u001b[39m])],\n\u001b[0;32m      <a href='vscode-notebook-cell:/f%3A/programming_projects/yield_prediction/yield_prediction.ipynb#ch0000020?line=3'>4</a>\u001b[0m     np\u001b[39m.\u001b[39margmax(cv_data[\u001b[39m'\u001b[39m\u001b[39mtest-Accuracy-mean\u001b[39m\u001b[39m'\u001b[39m])\n\u001b[0;32m      <a href='vscode-notebook-cell:/f%3A/programming_projects/yield_prediction/yield_prediction.ipynb#ch0000020?line=4'>5</a>\u001b[0m ))\n",
      "\u001b[1;31mNameError\u001b[0m: name 'cv_data' is not defined"
     ]
    }
   ],
   "source": [
    "print('Best validation accuracy score: {:.2f}±{:.2f} on step {}'.format(\n",
    "    np.max(cv_data['test-Accuracy-mean']),\n",
    "    cv_data['test-Accuracy-std'][np.argmax(cv_data['test-Accuracy-mean'])],\n",
    "    np.argmax(cv_data['test-Accuracy-mean'])\n",
    "))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 284,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precise validation accuracy score: 0.916256863932463\n"
     ]
    }
   ],
   "source": [
    "print('Precise validation accuracy score: {}'.format(np.max(cv_data['test-Accuracy-mean'])))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['accuracy', 'adjusted_mutual_info_score', 'adjusted_rand_score', 'average_precision', 'balanced_accuracy', 'completeness_score', 'explained_variance', 'f1', 'f1_macro', 'f1_micro', 'f1_samples', 'f1_weighted', 'fowlkes_mallows_score', 'homogeneity_score', 'jaccard', 'jaccard_macro', 'jaccard_micro', 'jaccard_samples', 'jaccard_weighted', 'max_error', 'mutual_info_score', 'neg_brier_score', 'neg_log_loss', 'neg_mean_absolute_error', 'neg_mean_absolute_percentage_error', 'neg_mean_gamma_deviance', 'neg_mean_poisson_deviance', 'neg_mean_squared_error', 'neg_mean_squared_log_error', 'neg_median_absolute_error', 'neg_root_mean_squared_error', 'normalized_mutual_info_score', 'precision', 'precision_macro', 'precision_micro', 'precision_samples', 'precision_weighted', 'r2', 'rand_score', 'recall', 'recall_macro', 'recall_micro', 'recall_samples', 'recall_weighted', 'roc_auc', 'roc_auc_ovo', 'roc_auc_ovo_weighted', 'roc_auc_ovr', 'roc_auc_ovr_weighted', 'top_k_accuracy', 'v_measure_score']\n"
     ]
    }
   ],
   "source": [
    "print(sorted(sklearn.metrics.SCORERS.keys()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Combating molecule sampling bias\n",
    "\n",
    "One way to overcome this bias is to only include one reaction per molecule in the dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "metadata": {},
   "outputs": [],
   "source": [
    "i, (c1, c2, p) = next(data[['Component_1', 'Component_2', 'Product']].iterrows())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'FC(F)(F)S(=O)(=O)OC1=CC2=C(C=CC=C2)C=C1'"
      ]
     },
     "execution_count": 126,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "c1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "54297it [00:42, 1271.42it/s]\n"
     ]
    }
   ],
   "source": [
    "molecules_to_reaction = defaultdict(lambda: defaultdict(set))\n",
    "for i, (c1, c2, p) in tqdm(data[['Component_1', 'Component_2', 'Product']].iterrows()):\n",
    "    label = y[i]\n",
    "    for smiles in (c1, c2, p):\n",
    "        smiles = canonicalize_smiles(record[col])\n",
    "        molecules_to_reaction[label][smiles].add(i)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 249,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "54297it [00:30, 1795.77it/s]\n"
     ]
    }
   ],
   "source": [
    "reaction_components = [(i, label, set([mol_id[canonicalize_smiles(smiles)] for smiles in smiles_list])) for i, (label, *smiles_list) in tqdm(data[['yieldclass', 'Component_1', 'Component_2', 'Product']].iterrows()) ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 271,
   "metadata": {},
   "outputs": [],
   "source": [
    "def split_train_test_reactions(reaction_components, test_proportion=1.0, rng=None):\n",
    "    if rng is None:\n",
    "        rng = np.random.default_rng()\n",
    "\n",
    "    chosen_molecules = set()\n",
    "    chosen_reactions = set()\n",
    "    chosen_labels = Counter()\n",
    "\n",
    "    rng = np.random.default_rng()\n",
    "\n",
    "    reaction_index = np.arange(len(reaction_components))\n",
    "    rng.shuffle(reaction_index)\n",
    "\n",
    "    class_remainders = defaultdict(list)\n",
    "    \n",
    "    for i in tqdm(reaction_index):\n",
    "        i, label, smiles_set = reaction_components[i]\n",
    "        if not smiles_set & chosen_molecules:\n",
    "            chosen_molecules.update(smiles_set)\n",
    "            chosen_reactions.add(i)\n",
    "            chosen_labels[label] += 1\n",
    "        else:\n",
    "            class_remainders[label].append(i)\n",
    "\n",
    "    test_reactions = set()\n",
    "    for label, count in chosen_labels.items():\n",
    "        n_reactions = int(test_proportion * count)\n",
    "        labeled_test_reactions = rng.choice(class_remainders[label], size=n_reactions, replace=False)\n",
    "        test_reactions.update(labeled_test_reactions)\n",
    "\n",
    "    return chosen_reactions, test_reactions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 283,
   "metadata": {},
   "outputs": [],
   "source": [
    "def make_symbolic_reactions(*reaction_dfs):\n",
    "    unique_molecules = set()\n",
    "    # First collect all molecules\n",
    "    for reaction_df in reaction_dfs:\n",
    "        for col in ('Component_1', 'Component_2', 'Product'):\n",
    "            unique_molecules.update(canonicalize_smiles(smiles) for smiles in reaction_df[col].unique())\n",
    "    mol_id = {smiles: i for smiles in enumerate(sorted(unique_molecules))}\n",
    "    return [reaction_df[['Component_1', 'Component_2', 'Product']].applymap(lambda x: mol_id[canonicalize_smiles(x)])]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 281,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['O=S(=O)(Oc1ccc2ccccc2c1)C(F)(F)F',\n",
       " 'Clc1cccc(I)c1',\n",
       " 'CSc1ccc(Br)cc1C#N',\n",
       " 'C[C@H](C(=O)O[C@@H](C(=O)N1CCOCC1)c1ccccc1)c1ccc(I)c(F)c1',\n",
       " 'CN(C)c1ccc(Br)cc1C(F)(F)F',\n",
       " 'O=C(O)Cc1cccc(Br)c1',\n",
       " 'Brc1cncnc1',\n",
       " 'BrCCCOc1ccccc1',\n",
       " 'Cc1ccc(S(=O)(=O)N(C(=O)c2ccccc2)c2ccccc2)cc1',\n",
       " 'CC(=O)Nc1ccc(F)cc1Br',\n",
       " 'COc1ccc(-c2ncc(I)c(-c3ccccc3)c2C(=O)c2ccccc2)cc1',\n",
       " 'O=Cc1ccc(Br)cc1',\n",
       " 'C[C@@H](Nc1ncnc2c1c(I)cn2COCC[Si](C)(C)C)c1ccccc1',\n",
       " 'N#Cc1ccc(I)cc1',\n",
       " 'Brc1ccc(-c2ccc(Br)s2)s1',\n",
       " 'Fc1ccc(Cl)cc1',\n",
       " 'CCOP(=O)(OCC)Oc1ccc2ccccc2c1',\n",
       " 'O=[N+]([O-])c1ccc(Br)cc1',\n",
       " 'CN(C)c1snc2cc(Br)cnc12',\n",
       " 'COc1ccc2c(c1)c(I)cn2S(=O)(=O)c1ccccc1',\n",
       " 'COC(=O)/C=C1\\\\NC(=O)C(C)=C1Br',\n",
       " 'Fc1ccc(Br)cc1',\n",
       " 'CCOC(=O)/C=C(/CNC(=O)OCc1ccccc1)OS(=O)(=O)c1ccc(C)cc1',\n",
       " 'CC/C(=C\\\\I)COCc1ccc(OC)cc1',\n",
       " 'CC(=O)Nc1nc2ccc(Br)cc2s1',\n",
       " 'NC(=O)[C@H]1CC[C@H](CNc2nc(Cl)cc(NCc3ccccc3)n2)CC1',\n",
       " 'CCCCCCCCC1(CCCCCCCC)c2ccccc2-c2ccc(Br)cc21',\n",
       " 'O=[N+]([O-])c1ccccc1Br',\n",
       " 'Cc1cc(=O)oc2ccc(OS(=O)(=O)C(F)(F)F)cc12',\n",
       " 'COC(=O)c1cc(C(C)(C)C)cc2c1Oc1c(Br)cc(C(C)(C)C)cc1C2(C)C',\n",
       " 'Cc1cccc(C#N)c1I',\n",
       " 'BrC(Br)=C(c1ccccc1)c1ccccc1',\n",
       " 'Nc1ccc(Cl)nc1I',\n",
       " 'COc1ccc(S(=O)(=O)n2ccc3cccc(Br)c32)cc1',\n",
       " 'COc1ccc(-c2sc(N3C(=O)c4ccccc4C3=O)c(C(=O)c3ccc(Cl)cc3)c2CBr)cc1',\n",
       " 'O=C(N1CCCCC1Cc1ccccc1)n1cc(-c2ccc(Br)cc2)nn1',\n",
       " 'Ic1c2nc(c(-c3ccccc3)c3ccc(cc4nc(c(-c5ccccc5)c5ccc1[nH]5)C=C4)[nH]3)C=C2',\n",
       " 'O=C(c1ccccc1)c1ccc(Cl)cc1',\n",
       " 'COc1ccc(CN(Cc2ccc(OC)cc2OC)S(=O)(=O)Oc2cccc(Br)c2)c(OC)c1',\n",
       " 'C[C@@H]1C(=O)C(C#N)=C[C@]2(c3ccccc3)c3nn(C)c(-c4cccc(Br)c4)c3CC[C@@H]12',\n",
       " 'Cc1ccc(S(=O)(=O)N(C(=O)c2ccc(NC(=O)OC(C)(C)C)cc2)c2cc(C(F)(F)F)cc(C(F)(F)F)c2)cc1',\n",
       " 'COCOc1cc(Br)c(Br)cc1OCOC',\n",
       " 'Nc1ncc(-c2ccc(CN3CCOCC3)s2)c2scc(Br)c12',\n",
       " 'CCCCCCn1c2ccc(Br)cc2c2cc([N+](=O)[O-])ccc21',\n",
       " 'COc1ccc(CN(C2CCCC2)S(=O)(=O)c2cc(Br)ccc2OC)c(OC)c1',\n",
       " 'Cc1ccc(S(=O)(=O)n2ccc3c(-c4cc(Br)ccc4OCC4CC4)cn(C)c(=O)c32)cc1',\n",
       " 'N#Cc1cccc(CNc2nc(C(F)(F)F)nc3ncc(Br)cc23)c1',\n",
       " 'N#Cc1ccccc1Br',\n",
       " 'C=CCCN(C=C(Cl)Cl)S(=O)(=O)c1ccc(C)cc1',\n",
       " 'Brc1cc2[nH]c3cc(Br)c(Br)cc3c2cc1Br',\n",
       " 'COc1cccc(-c2cc(Br)cs2)c1',\n",
       " 'OCc1ccccc1Cl',\n",
       " 'CC1(C)Cc2nc1cc1cc(C#Cc3ccccc3)c(cc3nc(cc4ccc([nH]4)c2Br)C=C3)[nH]1',\n",
       " 'COc1ccc2ccccc2c1Br',\n",
       " 'N#Cc1ccc(Br)cc1',\n",
       " 'Cc1ccc(S(=O)(=O)n2ccc3c(Cl)ncnc32)cc1',\n",
       " 'Ic1c(OCc2ccccc2)nc2ccccc2c1OCc1ccccc1',\n",
       " 'Clc1nccc2ccccc12',\n",
       " 'Cc1ccc(S(=O)(=O)Oc2ncc(-c3ccccc3)cc2-c2ccccc2)cc1',\n",
       " 'Clc1ncnc2ccsc12',\n",
       " 'Cc1ccc(C(=O)OC[C@H]2O[C@@H](c3ccc(Br)cc3)C[C@@H]2OC(=O)c2ccc(C)cc2)cc1',\n",
       " 'O=c1c(Br)nsnc1-c1ccccc1',\n",
       " 'Ic1c(-c2ccccc2)n2nc3c(c2c2ccccc12)CCCC3',\n",
       " 'Cc1c(-c2ccc(N3CCCC(O)C3)nc2)sc2c(N3CCOCC3)nc(Cl)nc12',\n",
       " 'COC(=O)c1c(OS(=O)(=O)C(F)(F)F)c(OS(=O)(=O)C(F)(F)F)c(C(=O)OC)n1CCc1ccc(OC)cc1',\n",
       " 'Brc1cccs1',\n",
       " 'C/C(=C/I)CO',\n",
       " 'COc1cc(-c2[nH]c3ccccc3c(=O)c2Br)cc(OC)c1OC',\n",
       " 'COP(=O)(OC)c1ccc2ccccc2c1Br',\n",
       " 'O=C(F)/C=C/c1ccccc1',\n",
       " 'N[C@@H](c1ccccc1)c1ccc(Br)cc1',\n",
       " 'COCn1c(-c2ccc(N3CCN(C(=O)OC(C)(C)C)CC3)cc2)c(I)c2c(Cl)ccnc21',\n",
       " 'Cc1ccc(-c2cc(-c3cccc(Br)c3)nc(-c3ccccc3O)c2)cc1',\n",
       " 'Nc1ncnc2c1nc(-c1ccc(Cl)cc1)n2[C@H]1C[C@H](O)[C@@H](CO)O1',\n",
       " 'COc1cc2c(c(Br)c1OC)CN(CCc1ccccc1)CC2',\n",
       " 'O=C(Cl)c1ccccc1',\n",
       " 'Clc1nc(-c2ccccc2)c(Cl)c(-c2ccccc2)n1',\n",
       " 'Cn1c2c(c3ccccc31)C1(CC=CC1)c1cc(Br)ccc1-2',\n",
       " 'Brc1ccc(Br)cc1',\n",
       " 'Cc1ccc(S(=O)(=O)N(C(=O)c2ccc(Br)cc2)c2ccccc2)cc1',\n",
       " 'Cn1c(Br)c(C=O)c2ccccc21',\n",
       " 'O=Cc1cc(F)ccc1Br',\n",
       " 'Cc1cc(C)c(CNC(=O)c2cc(Br)nc3c2cnn3C(C)c2ccccc2)c(=O)[nH]1',\n",
       " 'O=C1CCc2cc(Br)ccc21',\n",
       " 'O=c1cc(-c2ccccc2)oc2cc(Br)c(Br)cc12',\n",
       " 'Cc1ccc(S(=O)(=O)Oc2ccc(Cl)cc2)cc1',\n",
       " 'O=C(O)c1ccc(Cl)cc1',\n",
       " 'CS(=O)(=O)c1ccncc1Nc1cc(Cl)c2nccnc2c1',\n",
       " 'COc1ccc(CSc2ccccc2Br)cc1',\n",
       " 'C=C(OC/C=C(/C)I)C(=O)OC(C)(C)C',\n",
       " 'CN(Cc1cc(Br)n(S(=O)(=O)c2ccccc2)c1)C(=O)OC(C)(C)C',\n",
       " 'O=C(NCc1ccco1)c1cc(S(=O)(=O)N2CCCC2)c(Br)s1',\n",
       " 'CC(=O)c1ccc(OS(=O)(=O)C(F)(F)F)c2c1CCCC2',\n",
       " 'CCCc1ccc(-c2ccc(-c3ccc(I)cc3)cc2)c(F)c1F',\n",
       " 'Clc1nc(-c2ccccc2)cc(-c2ccccc2)n1',\n",
       " 'Cc1ccc(S(=O)(=O)N2C[C@H]2c2ccccc2)cc1',\n",
       " 'COC(=O)c1c(OS(=O)(=O)C(F)(F)F)cc(C)c(C)c1C',\n",
       " 'COc1cc(Br)c(OC)cc1Br',\n",
       " 'CN(C)S(=O)(=O)c1cccc(-c2ccnc3c(I)c(-c4ccncc4)nn23)c1',\n",
       " 'COc1c(NCC2CCCCC2)cnc2ccc(Br)cc12',\n",
       " 'Cn1c(=O)sc2ccc(Br)cc21',\n",
       " 'Cc1c(C)c(Br)c(C)c(C)c1O',\n",
       " 'Brc1c2ccccc2c(-c2c3ccccc3c(Br)c3ccccc23)c2ccccc12',\n",
       " 'Cn1nc(I)c2cccnc21',\n",
       " 'COc1cc(Br)c2c(c1)N(C)C(=O)C2',\n",
       " 'COc1ccc(C#Cc2ccc(Br)cc2)cc1',\n",
       " 'O=C(c1ccc(Br)s1)C(F)(F)F',\n",
       " 'Cc1ccc(O)c(Br)c1',\n",
       " 'Nc1ncc(Br)c(-c2c[nH]c3ccccc23)n1',\n",
       " 'COC(=O)C1=C(OS(=O)(=O)C(F)(F)F)CCCCC1',\n",
       " 'Ic1ccc(I)cc1',\n",
       " 'Clc1nccc2c1sc1ccccc12',\n",
       " 'CCCCCCc1cc(Br)sc1-c1nc2sc(-c3sc(Br)cc3CCCCCC)nc2s1',\n",
       " 'COc1ccc(C(=Cc2ccccc2)OC(C)=O)cc1',\n",
       " 'COc1ccc(I)cc1',\n",
       " 'CC(C)(C)OC(=O)N[C@@H](Cc1ccc(Br)cc1)C(=O)O',\n",
       " 'O=C(O)c1cccc(Br)c1',\n",
       " 'COC(=O)c1cc(Br)c(C(=O)OC)cc1Br',\n",
       " 'COc1ccc2ccc(Br)cc2c1-c1c(OC)ccc2ccc(Br)cc12',\n",
       " 'Brc1nccc2ccncc12',\n",
       " 'Nc1nc2c(Br)cccn2n1',\n",
       " 'O=S(=O)(Nc1ccccc1)c1ccc(Br)cc1',\n",
       " 'CCCCNC(=O)c1ccc(Br)cc1',\n",
       " 'CCOP(=O)(OCC)OC1=C(Br)CC1',\n",
       " 'C=C(Br)CCO',\n",
       " 'BrC(Br)=c1c2ccccc2c(=C(Br)Br)c2nc(-c3ccccc3)c(-c3ccccc3)nc12',\n",
       " 'CC(O[C@@H]1COc2nc([N+](=O)[O-])cn2C1)c1ccc(Br)cc1',\n",
       " 'COc1cc(OC)cc(OS(=O)(=O)c2c(F)c(F)c(F)c(F)c2F)c1',\n",
       " 'Cc1ccc(S(=O)(=O)Oc2ccc3ccc(OS(=O)(=O)c4ccc(C)cc4)cc3c2)cc1',\n",
       " 'N#Cc1ccc(CBr)cc1',\n",
       " 'C=CCC#CC(OC(=O)OC)c1ccc(Cl)cc1',\n",
       " 'Brc1csc(-c2cn([C@H]3CN4CCC3CC4)nn2)c1',\n",
       " 'Brc1ccc(-c2cc(-c3ccccn3)nc(-c3ccccn3)c2)cc1',\n",
       " 'COc1ccc2c(OS(=O)(=O)C(F)(F)F)cccc2c1',\n",
       " 'O=C1C(Br)=C(Br)C(=O)c2ccccc21',\n",
       " 'O=C(c1ccc(Br)cc1)c1cc2cc(Br)ccc2o1',\n",
       " 'Brc1sc(Br)c(Br)c1Br',\n",
       " 'CCOC(=O)C1CCCN(CCC#Cc2ccccc2Br)C1',\n",
       " 'Cn1cc(I)c(=O)c2ccccc21',\n",
       " 'Cc1cc(Oc2ccccc2)c(Cl)c(=O)o1',\n",
       " 'COc1ncc(I)c(OC)n1',\n",
       " 'CC1(C)CCN(CCOc2ccc3[nH]ccc3c2)c2ccc(Br)cc21',\n",
       " 'Nc1nc(N2CCOCC2)c2nc(Cl)ccc2n1',\n",
       " 'Clc1ccccc1-n1ccc2cccnc21',\n",
       " 'COC(=O)[C@]12CS(=O)(=O)[C@H](C(C)(C)C)N1C(=O)C=C2OS(C)(=O)=O',\n",
       " 'Fc1cc(Br)ccc1Cl',\n",
       " 'Clc1ccc(Cl)cc1',\n",
       " 'Oc1ccc2cc(Br)ccc2c1',\n",
       " 'Cn1c(Br)c(Br)c2ccccc21',\n",
       " 'O=Cc1cccc(Br)c1',\n",
       " 'O=C(Nc1ccc(Br)c2cccnc12)c1ccccc1',\n",
       " 'Cn1nnn(-c2cc([N+](=O)[O-])c(F)cc2Br)c1=O',\n",
       " 'O=Cc1cc2ccccc2nc1Cl',\n",
       " 'COc1cc(OC(=O)C(C)(C)C)cc(OS(=O)(=O)c2ccc(C)cc2)c1',\n",
       " 'CC(C)(C)c1cc(Br)c(I)c(Br)c1',\n",
       " 'C=C1CC(=O)N[C@@H](C)C(=O)N(C)[C@@H](C(C)C)C(=O)NCC(=O)O[C@@H](C(C)=CI)[C@@H]1C',\n",
       " 'Nc1nc(Cl)cc(N2CCCC2C2CC2)n1',\n",
       " 'CN1CCCN(C2CCC(Oc3cc(=O)n(C)cc3Br)CC2)C1=O',\n",
       " 'Brc1cnc2nnc(Cc3ccc4ncccc4c3)n2c1',\n",
       " 'Brc1ccc(Br)nc1',\n",
       " 'CCOC(=O)c1ccc(I)cc1',\n",
       " 'Clc1nc(Cl)nc(-c2ccccc2)n1',\n",
       " 'O=C1Nc2cccc(I)c2C1=O',\n",
       " 'COc1cc(C=C(F)Br)cc(OC)c1',\n",
       " 'Cc1nc(C)c([C@H](OC(C)(C)C)C(=O)OC(C)C)c(N2CCC3(CCC3)CC2)c1Br',\n",
       " 'CCc1cc2c(cc1O[Si](C)(C)C(C)(C)C)CC[C@@H]1[C@@H]2CC[C@]2(C)C(=O)CC[C@@H]12',\n",
       " 'CC(C)(C)c1cc(I)cc(I)c1',\n",
       " 'COCOC(=O)/C=C(\\\\I)CCO',\n",
       " 'CCOC(=O)c1cc(Br)c(C#N)[nH]1',\n",
       " 'Brc1nccs1',\n",
       " 'O=C1N(Cc2ccccc2)[C@H](c2ccccc2)[C@]1(F)Br',\n",
       " 'Brc1ccc2nc(OCc3cccc4ccccc34)c(I)cc2c1',\n",
       " 'FC(F)(F)c1cc(-c2cc([S+]3CCCC3)c3cccccc2-3)cc(C(F)(F)F)c1',\n",
       " 'O=C(c1ccccc1Br)c1ccccc1Br',\n",
       " 'CC(C)[Si](O[C@@H]1c2ccccc2C(OS(=O)(=O)C(F)(F)F)=C2COC[C@H]21)(C(C)C)C(C)C',\n",
       " 'CC(C)(C)C(=O)OCn1cnc2c(Cl)nc(I)nc21',\n",
       " 'Nc1ncn(Cc2ccccc2)c2nc(Br)nc1-2',\n",
       " 'Nc1ccccc1C(=O)c1ccc(Br)cc1',\n",
       " 'CC(/C=C(\\\\I)CO)CO[Si](c1ccccc1)(c1ccccc1)C(C)(C)C',\n",
       " 'COC(=O)c1ccc(Cl)cc1',\n",
       " 'C[C@H](NC(=O)OCc1ccccc1)c1nc2ccc(F)cn2c1I',\n",
       " 'NNc1ccccc1',\n",
       " 'CC(C)Nc1ncc(Br)n(CC(=O)OC(C)(C)C)c1=O',\n",
       " 'Clc1ccc(N2CCOCC2)nn1',\n",
       " 'NC(=O)Nc1sc(Br)cc1C(N)=O',\n",
       " 'COC(=O)C1=C(OS(=O)(=O)C(F)(F)F)CCC1',\n",
       " 'COc1cccc(Cl)c1',\n",
       " 'O=Cc1ccc(Br)s1',\n",
       " 'O=CC1=C(Cl)c2c(ccc3ccccc23)CC1',\n",
       " 'O=C(O)c1ccc(Br)cc1',\n",
       " 'CC(C)Oc1ccccc1Cl',\n",
       " 'CCCCCCCCCCCC(=O)Nc1ccc(Br)cc1',\n",
       " 'O=Cc1cc(-c2cccs2)ccc1OS(=O)(=O)C(F)(F)F',\n",
       " 'Brc1nc(-c2ccco2)n(Cc2ccccc2)c1Br',\n",
       " 'CCn1c(=O)c(Br)cn(-c2ccc(Cl)cc2)c1=O',\n",
       " 'BrC#Cc1ccccc1',\n",
       " 'CO/N=C(\\\\CCl)c1ccccc1',\n",
       " 'O=c1ccn(-c2cccc(C(F)(F)F)c2)nc1Br',\n",
       " 'C/C(=C\\\\[C@@H](OC(=O)C(C)(C)C)c1ccccc1)CCC1OC1(C)C',\n",
       " 'COc1ccc(-c2oc3cc(OC)cc(C=O)c3c2I)cc1',\n",
       " 'COCn1ncc(Cl)c(OC)c1=O',\n",
       " 'Cc1ccc(C)c(Br)c1',\n",
       " 'Brc1cccc(-c2cccc(-c3cccc(Br)n3)n2)n1',\n",
       " 'Clc1cc(NCc2ccccc2)ncn1',\n",
       " 'COc1cc2c(Cl)cc(Cl)nc2cc1OCCCN1CCOCC1',\n",
       " 'C[C@@H](Oc1cc(Br)cnc1N(C(=O)OC(C)(C)C)C(=O)OC(C)(C)C)c1c(Cl)ccc(F)c1Cl',\n",
       " 'Brc1cc(C=Cc2c3nc(c(-c4ccccc4)c4ccc(cc5nc(c(-c6ccccc6)c6ccc2[nH]6)C=C5)[nH]4)C=C3)c(Br)s1',\n",
       " 'N#Cc1cc(Br)ccc1N',\n",
       " 'CCCCOC(=O)c1sc2c(ccc3c(Cl)c(C(=O)OCCCC)sc32)c1Cl',\n",
       " 'O=C(Cc1c[nH]c2ccc(Br)cc12)NC12CC3CC(CC(C3)C1)C2',\n",
       " 'C=C(CBr)c1cccc(OC)c1',\n",
       " 'Nc1nccc(Cl)n1',\n",
       " 'CNc1nc2ccccc2n2c(Br)cnc12',\n",
       " 'Brc1cnc2ccccc2c1',\n",
       " 'COC1=C(Br)C(=O)C(OC)=C(c2ccc(OC)cc2)C1=O',\n",
       " 'CC(C)(C)OC(=O)N1CCC2(C=C(OS(=O)(=O)C(F)(F)F)c3ccccc3O2)CC1',\n",
       " 'COC(=O)C1=Cc2cc(Br)ccc2S(=O)(=O)CC1',\n",
       " 'O=C1c2ccccc2-c2[nH]nc(-c3ccc(Br)cc3)c21',\n",
       " 'CCCCCCOc1cc(Br)c(OCCCCCC)cc1Br',\n",
       " 'Nc1ccccc1Cl',\n",
       " 'Oc1c(F)cc(Br)cc1F',\n",
       " 'Cc1cc(Cl)n(-c2ccccc2)n1',\n",
       " 'COc1ccc(-c2ccc(OC)c(-c3ccccc3)c2)cc1Br',\n",
       " 'Brc1ccccc1-c1nnc(-c2ccccc2Br)o1',\n",
       " 'Clc1nc2ccccc2n2cccc12',\n",
       " 'Cc1ccc(Br)cc1',\n",
       " 'O=c1cc(N2CCOCC2)oc2cc(OS(=O)(=O)C(F)(F)F)ccc12',\n",
       " 'Ic1ccccc1',\n",
       " 'CCOC(=O)c1cc(-c2ccc(C)cc2)ccc1-c1ccc(-c2ccc(-c3cc(OC)cc(OS(=O)(=O)C(F)(F)F)c3)c(F)c2)c(OC)c1',\n",
       " 'Brc1cc(Br)c(Br)cc1Br',\n",
       " 'Cc1c(C)c(C)c(Br)c(C)c1C',\n",
       " 'COc1cc(C#N)cc(Cl)c1OC',\n",
       " 'O=[N+]([O-])c1ccc(O)c(I)c1',\n",
       " 'Cc1cccnc1Br',\n",
       " 'C[Si](C)(C)CCOCn1cc(C(=O)Nc2ccc3[nH]ncc3c2)c2ncc(Br)cc21',\n",
       " 'O=Cc1cc(-c2ccccc2)cnc1Cl',\n",
       " 'Cc1cnc(Cl)c2c1c1ccccc1n2C',\n",
       " 'CCn1nc(-c2cccc(Br)c2)cc(NC(=O)Nc2c(Cl)cncc2Cl)c1=O',\n",
       " 'O=Cc1ccccc1OS(=O)(=O)C(F)(F)F',\n",
       " 'Brc1c2ccccc2c(C(=C(c2ccccc2)c2ccccc2)c2ccccc2)c2ccccc12',\n",
       " 'Clc1cc(Cl)nc(Cl)n1',\n",
       " 'CC(=O)Nc1ncc(C)c(Br)n1',\n",
       " 'Brc1cnc(-c2ccc(N(c3ccccc3)c3ccccc3)cc2)c2nsnc12',\n",
       " 'Brc1cc2sc(Br)cc2s1',\n",
       " 'C=C(CBr)c1ccc(C#N)cc1',\n",
       " 'CC(C)c1nn(-c2ccc(C#N)c(N[C@H]3CC[C@H](O)CC3)c2)c2nccc(I)c12',\n",
       " 'Brc1ccccc1I',\n",
       " 'Cc1cc(Br)ccc1C#N',\n",
       " 'CCOC(=O)c1[nH]c(I)c2ccc3nsnc3c12',\n",
       " 'N#Cc1nc(-c2ccc(I)cc2)c(-c2ccc(I)cc2)nc1C#N',\n",
       " 'CC(C)(C)[Si](C)(C)OCCC[n+]1c(-c2ccccc2)cc(-c2ccccc2)cc1-c1ccccc1',\n",
       " 'Brc1ccc2cc[nH]c2c1',\n",
       " 'COC(=O)c1c(OS(=O)(=O)C(F)(F)F)c(OS(=O)(=O)C(F)(F)F)c(C(=O)OC)n1Cc1ccccc1',\n",
       " 'N#Cc1ccc(Cl)cc1',\n",
       " 'COc1nc(Cl)nc2c1ncn2[C@H]1C[C@H](O[Si](C)(C)C(C)(C)C)[C@H](O[Si](C)(C)C(C)(C)C)[C@@H](CO[Si](C)(C)C(C)(C)C)O1',\n",
       " 'O=C1CCCc2c(Br)cccc21',\n",
       " '[O-][n+]1c(Cl)c(OC2CCCC2)nc2ccc(F)cc21',\n",
       " 'CC(C)(C)C(O)(c1cncnc1)c1ccc(Br)cc1C(F)(F)F',\n",
       " 'Cc1ccc2nc(Cl)nc(NCCN(C)C)c2c1',\n",
       " 'Brc1ccc2[nH]c3ccc(Br)cc3c2c1',\n",
       " 'CCN(CC)C(=O)Oc1ccc(C#N)cc1',\n",
       " 'CC1(C)Cc2nc1cc1ccc(cc3nc(cc4ccc([nH]4)c2Br)C=C3)[nH]1',\n",
       " 'COc1cc2c(cc1Br)Cc1cc(OC)c(Br)cc1Cc1cc(OC)c(Br)cc1C2',\n",
       " 'CCCCC(CC)Cn1c2ccc3cccc4c5cccc6c(Br)cc1c(c65)c2c34',\n",
       " 'O=S(=O)(OC1=CCCCC1)C(F)(F)F',\n",
       " 'C[C@@H](Nc1ncnc2sc(Br)cc12)c1ccccc1',\n",
       " 'BrCc1ccccc1',\n",
       " 'CCOP(=O)(OCC)OC[C@H]1O[C@@H](n2cnc3c(Cl)ncnc32)[C@@H]2OC(C)(C)O[C@H]12',\n",
       " 'CC(C)(C)OC(=O)N1[C@H]2CC[C@@H]1C=C(OS(=O)(=O)C(F)(F)F)C2',\n",
       " 'Brc1ccc2ccccc2n1',\n",
       " 'CS(=O)(=O)OCc1ccccc1',\n",
       " 'O=C(Oc1ccccc1)c1ccccc1',\n",
       " 'CC(=O)c1ccc(Cl)cc1',\n",
       " 'COC(=O)c1sccc1Br',\n",
       " 'Cc1ccc(S(=O)(=O)OC2=CC(=O)N(C(=O)OC(C)(C)C)[C@H]2Cc2ccccc2)cc1',\n",
       " 'Cc1ccc(-c2ccccc2I)cc1',\n",
       " 'COc1ccc(CN2CC(OS(=O)(=O)C(F)(F)F)=C(c3ccccc3)C2=O)cc1OC',\n",
       " 'O=C(c1ccccc1)n1cccc1',\n",
       " 'Cc1nn(-c2c(F)cccc2F)c2c1cc(Br)c(=O)n2C',\n",
       " 'O=C1CNC(=O)/C(=C/c2c[nH]c3cc(Br)ccc23)N1',\n",
       " 'BrCCC1OCCCO1',\n",
       " 'Cc1nc(Cl)cc(Cl)n1',\n",
       " 'CC(C)(C)OC(=O)NCCOc1ccc(Br)cc1C=O',\n",
       " 'COc1ccc(I)cc1OCc1ccccc1',\n",
       " 'COc1nc(Cl)c(C=O)c(OC)n1',\n",
       " 'O=S(=O)(F)c1cncc(Br)c1',\n",
       " 'Cc1c(-c2ccc(Br)s2)sc(-c2ccc(Br)s2)c1C',\n",
       " 'C=C(OS(=O)(=O)C(F)(F)C(F)(F)C(F)(F)C(F)(F)F)P(=O)(OCC)OCC',\n",
       " 'CN(C)C(=O)OCc1ccccc1',\n",
       " 'O=[N+]([O-])c1cccc(Cl)c1',\n",
       " 'Brc1ccnc(-c2ccccc2)c1',\n",
       " 'O=[N+]([O-])c1ccc(Br)nc1',\n",
       " 'COc1ccc(-c2cc(OS(=O)(=O)C(F)(F)F)c3cccn3c2)cc1',\n",
       " 'CC(C)(c1cccc2ccccc12)S(=O)(=O)c1ccccc1',\n",
       " 'COc1ccc(C(=O)/C=C/c2ccccc2)cc1OC',\n",
       " 'O=C(CCl)OCc1ccccc1',\n",
       " 'CC(=O)Nc1c(Br)c(-c2ccccc2)nn1C',\n",
       " 'COC(=O)c1cn2ccccc2c1C(=O)OC',\n",
       " 'CC(=O)OC[C@H]1O[C@@H](n2cnc3c(Cl)nc(COC(=O)c4ccccc4)nc32)[C@H](OC(C)=O)[C@@H]1OC(C)=O',\n",
       " 'COC(=O)C(/C=C/c1ccccc1)(C(=O)OC(C)(C)C)n1cc(Br)ccc1=O',\n",
       " 'COc1ccc(Cn2cc(Br)c[n+]2[O-])cc1',\n",
       " 'CC(C)[C@@H]1CC[C@@H](C)C[C@H]1[P@]1(=O)c2ccccc2-c2cc(OS(=O)(=O)C(F)(F)F)ccc21',\n",
       " 'O=c1[nH]c2ncc(Br)nc2n1C1CCCCC1',\n",
       " 'CSc1nn2c(I)cccc2c1NC(=O)OC(C)(C)C',\n",
       " 'CC(=O)c1ccc(OS(=O)(=O)N(C)C)cc1',\n",
       " 'CCc1nc(N)nc(N)c1I',\n",
       " 'Cn1cc(Br)cn1',\n",
       " 'Fc1cccc2c1cc1c3nc(Cl)ccc3ncn21',\n",
       " 'CCCCCCCCOc1c(OCCCCCCCC)c(-c2ccc(Br)s2)c2nsnc2c1-c1ccc(Br)s1',\n",
       " 'Cc1cc2nc(N3CCNCC3)n(Cc3ccc(F)c(Cl)c3)c2cc1C',\n",
       " 'Cc1cccc(C)c1Cl',\n",
       " 'CC(C)(C)OC(=O)N1CCC(C)(Oc2cc(Br)cc3c2cnn3C2CCCCO2)CC1',\n",
       " 'CN(C)C(=O)OCc1ccc(F)cc1',\n",
       " 'C=CCN1CCc2cc(Br)ccc2C1=O',\n",
       " 'COc1cccc(OC)c1C(=O)O',\n",
       " 'OC12CC3CC(C1)CC(NCc1cc(Br)on1)(C3)C2',\n",
       " 'Cc1c(I)c(C)c(I)c(C)c1I',\n",
       " 'Cc1cc(-c2ccccc2C=C(Br)Br)cs1',\n",
       " 'CCCCC1=C(Br)C(C)(C)OP1(=O)OCC',\n",
       " 'COc1cc2c(c(OC)c1OC)C(=O)/C(=C/Br)O2',\n",
       " 'COc1cc2nc3c(cnc4ccc(Br)cc43)cc2c(OC)c1OC',\n",
       " 'Brc1cc2cc3c(cc2cc1Br)C1C=CC3c2cc3cc(Br)c(Br)cc3cc21',\n",
       " 'CCOC(=O)c1nc(Br)n(C)c1-c1ccccc1',\n",
       " 'O=S(=O)(Oc1ccc(Br)c2c1oc1ccccc12)C(F)(F)F',\n",
       " 'COC[C@@H]1CCCN1C(=O)c1cc(I)ccc1OC',\n",
       " 'CC(=C\\\\C(Br)=C\\\\[C@@H](C)[C@@H](/C=C/C(Br)=C/CCc1ccccc1)O[Si](C)(C)C(C)(C)C)/C=C(\\\\C)C(=O)OCC[Si](C)(C)C',\n",
       " 'Clc1cc(Cl)ncn1',\n",
       " 'O=c1nc(N2CCOCC2)oc2c(Br)c(OCc3ccccn3)ccc12',\n",
       " 'Brc1ccc(-c2ccccn2)cc1',\n",
       " 'Cc1nc2cc(Cl)ccc2s1',\n",
       " 'CC(O)c1cncc(Br)c1',\n",
       " 'Cc1ccc(I)cc1',\n",
       " 'Cc1cc(=O)oc2cc(OCc3cn(CCCNc4nc(Br)cn5ccnc45)nn3)ccc12',\n",
       " 'Cc1ccc(Br)cc1S(=O)(=O)NC1CCCC1',\n",
       " 'COc1cc(Br)c2ccccc2c1O[Si](C)(C)C(C)(C)C',\n",
       " 'CCCCCCCCCCCCCCCCCCOc1cc(COC(=O)c2ccc(I)cc2)cc(OCCCCCCCCCCCCCCCCCC)c1OCCCCCCCCCCCCCCCCCC',\n",
       " 'CC1(c2ccccc2)CCN(c2cccc(Br)c2)C(=O)O1',\n",
       " '[O-][n+]1cc(Br)cc2ccccc21',\n",
       " 'CC1(C)C=Cc2c(ccc([N+](=O)[O-])c2OS(=O)(=O)C(F)(F)F)O1',\n",
       " 'Brc1ccc(Br)s1',\n",
       " 'Brc1c2c(nc3ccccc13)CCCC2',\n",
       " 'O=[N+]([O-])c1ccc(C#CI)cc1',\n",
       " 'Oc1ccc(I)cc1',\n",
       " 'C=C(CBr)c1ccccc1',\n",
       " 'Brc1ccc(-c2ccc(-c3cccs3)s2)s1',\n",
       " 'N#Cc1c(Br)[nH]c2nc(N)[nH]c(=O)c12',\n",
       " 'Ic1csc(-c2ccccc2)n1',\n",
       " 'CN1C(=O)CCc2cc(Br)cc(F)c21',\n",
       " 'Brc1csc(Br)n1',\n",
       " 'BrC1=Cc2c(ccc3cccc(-c4ccccc4)c23)OC1(c1ccccc1)c1ccccc1',\n",
       " 'Nc1ncc(Br)cc1[N+](=O)[O-]',\n",
       " 'Cc1sc(Cl)cc1C1=C(c2cc(Cl)sc2C)C(F)(F)C(F)(F)C1(F)F',\n",
       " 'CC1(C)c2c(Br)ccc(Br)c2C(C)(C)N1[O]',\n",
       " 'CCOC(=O)c1nc2ccccc2c(Br)c1C',\n",
       " 'C=C(Br)C(=O)OCC',\n",
       " 'Cc1cccc(C)c1Br',\n",
       " 'Brc1ccnc(-c2cc(Br)ccn2)c1',\n",
       " 'CC(C)Oc1ccc(Oc2ncc(Br)s2)cc1',\n",
       " 'Brc1ccccc1Br',\n",
       " 'O=C(C(=O)c1ccc(Br)cc1)c1ccc(Br)cc1',\n",
       " 'O=[N+]([O-])c1ccc(Br)c(F)c1',\n",
       " 'Brc1ccccn1',\n",
       " 'CN1OC[C@@H]2CN[C@H](c3ccc(Br)cc3)C[C@@H]21',\n",
       " 'COc1ccc(OC)c(Br)c1',\n",
       " 'Brc1ccncc1',\n",
       " 'Brc1ccc(Br)c2snnc12',\n",
       " 'Sc1c(Br)cc(Br)cc1Br',\n",
       " 'C[Si](C)(C)c1c(Br)cccc1Br',\n",
       " 'Brc1cnc2c(ccc3cc(Br)cnc32)c1',\n",
       " 'C=CCOC1(C)CCN(c2c([C@H](OC(C)(C)C)C(=O)OCC)c(C)nc3cc(-c4cccc(Br)c4)nn23)CC1',\n",
       " 'Cc1cccc(Cl)c1S(=O)(=O)Nc1cc(Br)cnc1Cl',\n",
       " 'CC(C)(C)c1ccc(Br)cc1',\n",
       " 'Brc1nnc(SCc2ccccc2)n1Cc1ccccc1',\n",
       " 'CC(=O)Oc1ccc([C@H](OC(C)=O)[C@H]2[C@@H](/C=C\\\\I)C(C)=C[C@@H]3[C@@H]2C(C)(C)C[C@@]3(C)O)cc1',\n",
       " 'CC(C)(C)[Si](C)(C)Oc1ccc(Nc2ccc(Br)cn2)cc1',\n",
       " 'Ic1nccn1-c1ccccc1-n1ccnc1I',\n",
       " 'O=Cc1cc(Br)c(-c2cc(Br)c(Br)s2)s1',\n",
       " 'COc1ccccc1Br',\n",
       " 'Clc1ncccn1',\n",
       " 'Nc1ccc(-c2ccccc2)cc1Br',\n",
       " 'CCOC(=O)c1c(CC)c2cc3nc(c(OC)c4[nH]c(cc5nc(c(Br)c1[nH]2)CC5(C)C)c(CC)c4C(=O)OCC)CC3(C)C',\n",
       " 'CCCCC#CC(OC(=O)OC)c1ccccc1',\n",
       " 'C[C@@H]1ON(C)[C@H]2C[C@H]3c4cc(Br)ccc4CCN3C(=O)[C@@H]12',\n",
       " 'O=[N+]([O-])c1ccc2c(I)nn(C3CCCCO3)c2c1',\n",
       " 'CC(C)[C@H]1NC(=O)C(I)=CN1C(=O)c1ccccc1',\n",
       " 'OC[C@H]1[C@H](I)[C@@H]1CCc1ccccc1',\n",
       " 'COc1ccc2cc(Br)ccc2c1',\n",
       " 'Clc1cc(NC2CCOC2)ncn1',\n",
       " 'COc1ccc(N(c2ccc(Br)cc2)c2ccc(C=Cc3ccc(N(c4ccc(Br)cc4)c4ccc(OC)cc4)cc3)cc2)cc1',\n",
       " 'Nc1ccc(Br)cn1',\n",
       " 'CCC1(CC)c2ccccc2-c2ccc(Br)cc21',\n",
       " 'COc1ccc(CN2CC(c3ccc(OC)cc3)=C(Br)C2=O)cc1',\n",
       " 'Cc1ccc(S(=O)(=O)Oc2ccccc2C)cc1',\n",
       " 'CC(C)c1cccc(C(C)C)c1Nc1cccc(Br)n1',\n",
       " 'O=[N+]([O-])c1ccc(Br)s1',\n",
       " 'OCc1c(F)cccc1Br',\n",
       " 'CCS/C(Cl)=C\\\\c1ccccc1',\n",
       " 'Brc1c(-c2ccccc2)nc2ccccn12',\n",
       " 'Cc1cccc(Cl)n1',\n",
       " 'FC(F)(F)c1ccc(Br)cc1',\n",
       " 'Brc1cccnc1',\n",
       " 'Cn1c(=O)n(C)c2c3cc(I)ccc3ncc21',\n",
       " 'Cn1nc2c(Br)ccc(Br)c2n1',\n",
       " 'CCOC(=O)c1[nH]c2ccc(CC)cc2c1I',\n",
       " 'CC(C)(C)OC(=O)/C=C/c1cc(Br)ccc1NC(=O)CCc1ccccc1',\n",
       " 'Clc1cn2c(n1)CCC2',\n",
       " 'O=C(/C=C/c1ccccc1)N1C(=O)c2ccccc2S1(=O)=O',\n",
       " 'CN(Cc1cc(Br)cc(Br)c1N)C1CCCCC1',\n",
       " 'O=C(O)Cc1cc(I)ccc1C(=O)O',\n",
       " 'Clc1ccc(-n2cccc2)cc1',\n",
       " 'O=CC1=C(Cl)c2cc3ccc4cccc5ccc(c2CC1)c3c45',\n",
       " 'O=C(c1ccccc1)c1ccc(Br)cc1',\n",
       " 'Cc1c(Br)cccc1[N+](=O)[O-]',\n",
       " 'Cc1ccc(S(=O)(=O)OC=C(F)F)cc1',\n",
       " 'Brc1ccoc1',\n",
       " 'O=[N+]([O-])c1cn2c(n1)OC[C@@H](OCc1ccc(Br)nc1)C2',\n",
       " 'CC(C)(C)c1ccc(I)cc1',\n",
       " 'CC(=O)Oc1cccc2c1C(=O)c1c(ccc(I)c1O)C2=O',\n",
       " 'COCCOCOC(=O)c1cc(OS(=O)(=O)C(F)(F)F)ccc1-c1ccc(C(=O)NCC(C)C)cc1C(=O)OCc1ccccc1',\n",
       " 'CCCCCCc1cc(-c2ccc(-c3cccs3)c3nsnc23)sc1Br',\n",
       " 'O=C(c1ccc(Br)cc1)N1CCc2c([nH]c3ccccc23)C1',\n",
       " 'COCOc1cc(C(C)C)c(Br)c(C(C)C)c1',\n",
       " 'N#Cc1ccccc1Cl',\n",
       " 'Cc1ccccc1Br',\n",
       " 'CS(=O)(=O)c1ccc(-c2cnc3ccc(Cl)nn23)cc1',\n",
       " 'N#CCC(=O)c1ccc(Br)cc1',\n",
       " 'Clc1nc(N2CCOCC2)c2ccsc2n1',\n",
       " 'Cc1ccc(S(=O)(=O)N(Cc2ccccc2)/C(Cl)=C/CO)cc1',\n",
       " 'Brc1ncn(Cc2ccccc2)c1-c1ccc2ccccc2c1',\n",
       " 'Brc1cc2c(s1)-c1sc(Br)cc1[Si]2(c1ccccc1)c1ccccc1',\n",
       " 'CC(C)(C)OC(=O)c1cccc(Br)n1',\n",
       " 'Cc1nn(C)c(-c2ccccc2)c1Br',\n",
       " 'Clc1ccc(Cl)nc1',\n",
       " 'CN(C(=O)Cn1c(=O)oc2ccc(Br)cc21)c1ccccc1',\n",
       " 'CSc1nccc(Br)n1',\n",
       " 'O=C(O)C(c1ccc(Cl)cc1)c1ccc(Cl)cc1',\n",
       " 'CC(C)c1ccc2cc(N(C)C)ccc2c1',\n",
       " 'COc1ccc(C(=O)N2C(=O)CCC2=O)cc1',\n",
       " 'Brc1ccc2cccc(Br)c2n1',\n",
       " 'Clc1cc(Cl)c(Cl)cc1Cl',\n",
       " 'Clc1cnc(-c2nc3ccccc3s2)c(Br)c1',\n",
       " 'COC(=O)c1ccc(OS(=O)(=O)c2ccc(C)cc2)cc1',\n",
       " 'CC(C)(C)OC(=O)N1CCC(C#N)(c2cccc(Br)c2)CC1',\n",
       " 'Cc1ccc(S(=O)(=O)OC(c2ccc(-c3ccccc3)cc2)C(F)(F)F)cc1',\n",
       " 'CCCc1c(F)cc(-c2ccc(-c3ccc(I)cc3)cc2)cc1F',\n",
       " 'CC(Cn1nc(I)c2c(N)ncnc21)c1ccccc1',\n",
       " 'Cc1ccc(S(=O)(=O)N(C)C(=O)c2ccc(C(F)(F)F)cc2)cc1',\n",
       " 'CC(C)(C)OC(=O)N1[C@H]2CC[C@@H]1C[C@H]2c1cnc(N)c(Br)c1',\n",
       " 'COC(=O)[C@@H]1CC(OS(=O)(=O)C(F)(F)F)=CN1C(=O)c1cc(OC)c(OCCCOc2cc([N+](=O)[O-])c(C(=O)N3C=C(OS(=O)(=O)C(F)(F)F)C[C@H]3C(=O)OC)cc2OC)cc1[N+](=O)[O-]',\n",
       " 'CSc1ccc(I)cc1',\n",
       " 'COc1ccc(C(C(=O)NC2CCCCC2)N2C(=O)C(I)=C(c3ccccc3)C23C=CC(=O)C=C3)c(C)c1',\n",
       " 'Brc1cc(Br)c2ccc3c(Br)cc(Br)c4ccc1c2c43',\n",
       " 'CCCc1ccc(-c2ccc(-c3ccc(I)c(F)c3F)cc2F)c(F)c1F',\n",
       " 'CC(Cn1ncc([N+](=O)[O-])c1Cl)N1CCOCC1',\n",
       " 'Fc1cccc(Br)c1',\n",
       " 'O=C(Cc1ccc(Cl)nc1)NCc1cccc(F)c1',\n",
       " 'Cn1c2ccccc2c(=O)c2c(Br)cccc21',\n",
       " 'N#Cc1ccccc1CBr',\n",
       " 'CCCCCC/C(Br)=C\\\\CCC(C)=O',\n",
       " 'Cn1ncc(Br)c1Nc1ccc(I)cc1',\n",
       " 'CS(=O)(=O)N1CCN(c2ccc(Nc3nc(Cl)cc(N4CCOCC4)n3)cc2)CC1',\n",
       " 'CC(C)c1cc(C(C)C)c(S(=O)(=O)Oc2nc(N)nc3c2ncn3[C@H]2C[C@H](O[Si](C)(C)C(C)(C)C)[C@@H](CO[Si](C)(C)C(C)(C)C)O2)c(C(C)C)c1',\n",
       " 'O=c1c(Cl)c2oc3ccc4ccccc4c3nc-2c2ccccc12',\n",
       " 'COc1ccc(Cl)cc1',\n",
       " 'Clc1cc(NCCCCN2CCOCC2)nc(-c2ccncc2)n1',\n",
       " 'CC1(C)OC(=O)c2c(I)cccc2S1',\n",
       " 'CN(C)c1ccc(Br)cc1',\n",
       " 'Brc1cnc2[nH]ccc2n1',\n",
       " 'CCCCn1c2ccc(Br)cc2c2cc(Br)ccc21',\n",
       " 'O=c1ccnc2cc(Br)ccn12',\n",
       " 'COc1ccc2cc(Br)c(=O)oc2c1',\n",
       " 'Clc1ncnc2ccccc12',\n",
       " 'COc1ccc(OS(=O)(=O)C(F)(F)F)cc1',\n",
       " 'CCCCCn1cc(C(=O)C(=O)NC23CC4CC(CC(C4)C2)C3)c2cc(Br)ccc21',\n",
       " 'O=c1cc(-c2ccccc2)oc2cc(OS(=O)(=O)C(F)(F)F)cc(OS(=O)(=O)C(F)(F)F)c12',\n",
       " 'O=C1c2ccccc2C(=O)C1(Cc1cccc(Br)c1)Cc1cccc(Br)c1',\n",
       " 'O=Cc1cc(Br)cc(Br)c1',\n",
       " 'CNC1=Nc2ccccc2N2C1=NCC2Br',\n",
       " 'CN1C(=O)Cc2c(Br)ccc(F)c21',\n",
       " 'CCN(CC)C(=O)c1cc(Br)ccc1N(C)C',\n",
       " 'CN1C(=O)[C@@]2(C[C@H]2c2ccc3c(I)n[nH]c3c2)c2ccccc21',\n",
       " 'Clc1nccn2cccc12',\n",
       " 'Brc1cccc2c1-c1ccccc1-2',\n",
       " 'N#CC(=Cc1ccccc1)CBr',\n",
       " 'C=CC[C@]12C(=O)C[C@H](C)C[C@H]1CC=C2I',\n",
       " 'Cl/C=C/C#Cc1ccccc1',\n",
       " 'ClC1=CCCC1',\n",
       " 'OC[C@@H](Nc1ncnc2sc(Br)cc12)c1ccccc1',\n",
       " 'Clc1ccc2cccc(Cl)c2n1',\n",
       " 'CCCCCN1C(=O)C2(COc3ncc(Br)cc32)c2cc(Br)ccc21',\n",
       " 'CC/C(=C(\\\\OS(=O)(=O)c1ccc(C)cc1)c1ccc(OC)cc1)c1ccccc1',\n",
       " 'CCOc1cc(C)[nH]c(=O)c1CNC(=O)c1cc(Cl)cc2c(C(C)C)ncn12',\n",
       " 'COCOc1cc(OC)c(Br)cc1C=O',\n",
       " 'O=C1C2(CC=CC2)Cc2cc(Br)ccc2C12CC=CC2',\n",
       " 'CC(C)NCCCn1c(Sc2cc3c(cc2I)OCO3)nc2c(N)ncnc21',\n",
       " 'N#Cc1ccc(F)c(Br)c1',\n",
       " 'C=CCCl',\n",
       " 'FC(F)(F)C(F)(F)C(F)(F)C(F)(F)C(F)(F)C(F)(F)CCI',\n",
       " 'O=c1cc(Cl)oc(N2CCOCC2)c1',\n",
       " 'Cc1ncc(I)[nH]1',\n",
       " 'O=C1c2cc(Br)ccc2S(=O)(=O)c2cc(Br)ccc21',\n",
       " 'Cc1ccnc(Cl)n1',\n",
       " '[O-][n+]1nc2ccnn2c2cc(I)ccc21',\n",
       " 'OC[C@H]1O[C@@H](n2cnc3c(Cl)ncnc32)[C@H](O)[C@@H]1O',\n",
       " 'O=S(=O)(F)Oc1cccc(Br)n1',\n",
       " 'CCCCCCCCC1(CCCCCCCC)c2cc(Br)ccc2-c2ccc(Br)cc21',\n",
       " 'C=CC/C(=C(\\\\Cl)c1ccccc1)S(=O)(=O)c1ccc(C)cc1',\n",
       " 'C=C(OC/C=C(\\\\C)I)C(=O)OC(C)(C)C',\n",
       " 'CN(C(=O)c1ccc2ncc(Br)n2c1)c1ccc(C#N)cc1',\n",
       " 'C[C@H]([C@H]1CC=C(CO)C(=O)O1)[C@H]1CC[C@H]2CC(OS(=O)(=O)C(F)(F)F)=CC[C@@]21C',\n",
       " 'Clc1nc(Cl)nc(N2CCOCC2)n1',\n",
       " 'Brc1nc(-c2ccccc2)cc(-c2ccccc2)n1',\n",
       " 'CN(C)c1ncnc2c1c(I)cn2[C@@H]1O[C@H](CO)[C@@H](O)[C@H]1O',\n",
       " 'Brc1ccc(-c2nc(-c3ccccn3)n(-c3ccccc3)n2)cc1',\n",
       " 'O=C(NCc1ccccc1)c1ccc(Br)cc1OCc1ccccc1',\n",
       " 'IC1=C(I)C(c2ccccc2)(c2ccccc2)Oc2ccccc21',\n",
       " 'O=c1occcc1OS(=O)(=O)C(F)(F)F',\n",
       " 'Cn1ccc2c3c(Cl)ncnc3n([C@@H]3O[C@H](COC(=O)c4ccccc4)[C@@H](OC(=O)c4ccccc4)[C@H]3OC(=O)c3ccccc3)c21',\n",
       " 'CN(C)CCCC1(c2ccc(F)cc2)OCc2cc(Br)ccc21',\n",
       " 'COc1ccc2cc(C3=C(I)C(=O)C4(C)C=CC=CN34)ccc2c1',\n",
       " 'OCCOCCOCCOc1cc(Br)cc(OCCOCCOCCO)c1OCCOCCOCCO',\n",
       " 'Cc1c(C(=O)NCCOc2ccccc2)cnc2c(Br)cnn12',\n",
       " 'C[C@@H](N1Cc2cc(Br)sc2C1=O)C(O)(Cn1cncn1)c1ccc(F)cc1F',\n",
       " 'C[C@]12C[C@@](C)(C#Cc3ccc(Br)cc3)C[C@](C)(C1)C(=O)NC2=O',\n",
       " 'CC(=O)c1ccc(Br)c(Cl)c1O',\n",
       " 'COc1ccc(-c2noc(-c3ccc(OC)cc3)c2I)cc1',\n",
       " 'O=C1CN(C(=O)c2cc(-c3cccc(Cl)c3)c(Br)o2)CN1',\n",
       " 'Cc1cc(-c2ncc(-c3cccnc3)cc2/C=C/c2ccccc2)cnc1Br',\n",
       " 'COC(=O)Oc1ccc2ccccc2c1',\n",
       " 'Brc1ccc(N(c2ccccc2)c2ccc(-c3ccccc3)cc2)cc1',\n",
       " 'CCOC(=O)C1=C(OS(=O)(=O)C(F)(F)F)CN(C(=O)OC(C)(C)C)CC1',\n",
       " 'Cc1cc2nc3c(nc2cc1C)Cn1c(=O)ccc2c(Br)cc(Br)c(c21)OC3',\n",
       " 'Cc1cccc(C)c1Nc1ncc(Br)n2cncc12',\n",
       " 'COC(=O)c1c(Br)sc2c(C3CC3)c(Cc3cccc4ccccc34)cc(=O)n12',\n",
       " 'Cn1cc(-c2cc3c(Cl)ncnc3[nH]2)cn1',\n",
       " 'CCCCC(Cl)CN(C)c1cccc(OC)c1',\n",
       " 'O=c1[nH]c(Cc2ccccc2)ncc1Br',\n",
       " 'COc1cc(C=O)c(Br)c(OC)c1OC',\n",
       " 'COc1ccc(C(OC(=O)C(C)(C)C)c2cccc3ccccc23)cc1',\n",
       " 'C=C(CBr)c1ccc2ccccc2c1',\n",
       " 'O=Cc1cc(Br)co1',\n",
       " 'O=C(O)c1nc(Br)cs1',\n",
       " 'Cc1cccc(NC(=O)N(C)C)c1',\n",
       " 'Cc1cc(I)ccc1NN',\n",
       " 'CCCCCCc1cc(Br)sc1-c1ccc(C2=C3C(=O)N(CC(CC)CCCC)C(c4ccc(-c5sc(Br)cc5CCCCCC)s4)=C3C(=O)N2CC(CC)CCCC)s1',\n",
       " 'Clc1ccn2c(n1)nc1ccccc12',\n",
       " 'COC(=O)c1ccc(I)cc1',\n",
       " 'CC(=O)C(=O)c1ccc(Cl)cc1',\n",
       " 'Fc1c(F)c(Br)c2nsnc2c1Br',\n",
       " 'CC(=O)c1ccc(Cl)s1',\n",
       " 'Nc1ccc(I)cc1',\n",
       " 'FC(F)(F)c1ccc(-c2ccc(/C(=C/c3ccc(Br)cc3)c3ccccc3)cc2)cc1',\n",
       " 'CCOC(=O)N1[C@@H]2CC=C(OS(=O)(=O)C(F)(F)F)[C@H]1CC2',\n",
       " 'COC(=O)c1cc([N+](=O)[O-])ccc1Br',\n",
       " 'O=S(=O)(NCCc1cccc(F)c1)c1cc(Br)cc(C(F)(F)F)c1',\n",
       " 'Ic1ccc(-c2ccc(-c3ccc(I)cc3)cc2)cc1',\n",
       " 'CCCCC(CC)CN1C(=O)/C(=C2/C(=O)N(CC(CC)CCCC)c3cc(Br)sc32)c2sc(Br)cc21',\n",
       " 'NC(=O)c1onc(OCc2ccccc2)c1I',\n",
       " 'Clc1ccccc1I',\n",
       " 'C[C@H]1OC(C)(C)O[C@H]1/C=C\\\\Br',\n",
       " 'CCOC(=O)c1nc(Br)n(-c2ccccc2)c1C',\n",
       " 'O=[N+]([O-])c1cc(Br)cnc1Nc1ccccc1',\n",
       " 'COC(=O)c1cc(OS(=O)(=O)C(F)(F)F)ccc1OS(=O)(=O)C(F)(F)F',\n",
       " 'N#Cc1ccc(OS(=O)(=O)F)cc1',\n",
       " 'O=C1c2c(Br)cccc2[C@@H]2C[C@H]1c1cccc(Br)c1C2=O',\n",
       " 'O=S(=O)(Oc1cccc(Cl)c1)C(F)(F)F',\n",
       " 'COc1cccc(C(=O)c2ccc(Br)cc2)c1',\n",
       " 'Nc1cncc(Br)c1',\n",
       " 'Nc1noc2cccc(I)c12',\n",
       " 'O=C(OCc1ccccc1)c1ccc(C(=O)Oc2ccccc2)cc1',\n",
       " 'O=c1ccc2c(Cl)nc(NC(CO)CO)nc2n1-c1c(F)cccc1F',\n",
       " 'COC(=O)c1cc(Br)c(Cl)cc1N',\n",
       " 'CCc1cc2cc(Br)ccc2nc1C',\n",
       " 'COc1ccc(-c2ccnc(-c3cccc(-c4cc(Cl)ccn4)n3)c2)cc1',\n",
       " 'O=[N+]([O-])c1cn2c(n1)OC[C@@H](Oc1ccc(Br)cn1)C2',\n",
       " 'Cc1ccc(S(=O)(=O)N2C=C(Br)[C@H]3CC[C@@H]2C3)cc1',\n",
       " 'CCCCCCOc1cc(N=Nc2cc(OCCCCCC)c(I)cc2OCCCCCC)c(OCCCCCC)cc1Br',\n",
       " 'FC(F)(F)c1nc(Cl)c2nc3ccccn3c2n1',\n",
       " 'COc1ccccc1I',\n",
       " 'Brc1ccc(Br)c2nsnc12',\n",
       " 'O=C(c1ccccc1)c1ccc(O)cc1',\n",
       " 'CC(C)(C)NC(=O)Nc1nc(Cl)c(-c2ccccc2Br)n(CC(=O)Nc2ccccc2C(=O)NS(=O)(=O)c2ccc(C(F)(F)F)cc2)c1=O',\n",
       " 'COc1ccc(CN2C(=O)C(Br)=C(c3ccc(C)cc3)C2=O)cc1',\n",
       " 'C[C@]12CC[C@H]3[C@@H](CC=C4CCCC[C@@]43C)[C@@H]1CC=C2I',\n",
       " 'C=C(Br)P(=O)(OCC)OCC',\n",
       " 'O=C1CC2CCCCN2C=C1I',\n",
       " 'Clc1ccc(Br)cc1',\n",
       " 'CCCCCCc1nc(-c2nc(CCCCCC)c(-c3ccc(C=O)s3)s2)sc1-c1ccc(Br)s1',\n",
       " 'COc1ccc(Cl)c([N+](=O)[O-])c1',\n",
       " 'CCN(CC)CCCNc1nc(Cl)c2c(n1)N(c1c(F)cccc1F)C(=O)NC2',\n",
       " 'C[C@]12CC[C@@H]3c4ccc(O)c(Br)c4CC[C@H]3[C@@H]1CCC2=O',\n",
       " 'COC(=O)/C(Br)=C1\\\\CCCO1',\n",
       " 'O=Cc1ccc(-c2ccc(Br)cc2)cc1',\n",
       " 'CSc1nc2c(nn3ccccc23)c(=O)[nH]1',\n",
       " 'O=Cc1cc(Br)c(C=O)cc1Br',\n",
       " 'CCCCCCCCCCC(CCCCCCCC)CN1C(=O)c2cc(Br)c3c4c(cc(Br)c(c24)C1=O)C(=O)N(CC(CCCCCCCC)CCCCCCCCCC)C3=O',\n",
       " 'COC(=O)c1ccc(CN2C[C@H](C)N(C(=O)c3cccc(I)c3)[C@H](C)C2)cc1',\n",
       " 'Brc1ccccc1-c1nc2ccccc2o1',\n",
       " 'COc1ccc(C#Cc2cc(Br)ccc2OC)cc1',\n",
       " 'Cc1ccc([S@](=O)N[C@@H](c2cccc(Br)n2)C(C)(C)C)cc1',\n",
       " 'COc1cc2nc(N)n[n+]([O-])c2cc1Br',\n",
       " 'Cc1cccc(C)c1OS(=O)(=O)C(F)(F)F',\n",
       " 'CC(=O)c1ccc(Br)nc1',\n",
       " 'Cc1nc(C)c(-c2ccc(Cl)nc2)s1',\n",
       " 'O=Cc1sccc1Br',\n",
       " 'C=C(Br)/C(C)=C1\\\\C(=O)N(C)c2ccccc21',\n",
       " 'COc1cc(Cc2cnc(N)nc2N)c(Br)c(OC)c1OC',\n",
       " 'CC(C)(C)OC(=O)N1CC[C@H](Oc2ccc3nccc(Br)c3c2)C1',\n",
       " 'COc1ccc(-c2oc3ccc(Br)cc3c(=O)c2O)cc1OC',\n",
       " 'COc1cc2c(cc1Br)Cc1c3c(c4c(c1-2)-c1cc(OC)c(Br)cc1C4)C(=O)N(CCC(C)C)C3=O',\n",
       " 'Cn1c(-c2ccccc2)nc2ccc(I)cc2c1=O',\n",
       " 'Cc1cccc(C)c1OS(=O)(=O)N(C)C',\n",
       " 'COC(=O)c1nc(Cl)c(N2CCCCCC2)nc1N',\n",
       " 'Cc1ccccc1Cl',\n",
       " 'COC(=O)CCC[C@H](O[Si](C)(C)C(C)(C)C)[C@@H](/C=C/I)O[Si](C)(C)C(C)(C)C',\n",
       " 'CCN(CC)N=Nc1cc(Br)c(C#Cc2ccccc2)c(Br)c1',\n",
       " 'Brc1ccc(I)cc1',\n",
       " 'COC(OC)c1ccccc1',\n",
       " 'OC[C@@H]1[C@@H](O)[C@H](O)[C@@H](O)CN1CCCCCOCc1ccc(Br)cc1',\n",
       " 'COCOc1c(-c2ccccc2)cc(OCc2ccccc2)c(Br)c1OCc1ccccc1',\n",
       " 'Cc1c(Br)cc(C(C)(C)C)c(O)c1-c1c(C)c(Br)cc(C(C)(C)C)c1O',\n",
       " 'O=S(=O)(Nc1ccc2[nH]nc(I)c2c1)c1ccc(-c2ccccc2)cc1',\n",
       " 'CC(C)(C)OC(=O)N(Cc1ccccn1)S(=O)(=O)c1ccc(Br)c2ccccc12',\n",
       " 'COc1ccc(-c2cc3c(Cl)ncnc3n2C)cc1',\n",
       " 'C=Cc1cc2ccc(Br)cc2[nH]c1=O',\n",
       " 'Cl[Se](Cl)(Cl)c1ccccc1',\n",
       " 'N#Cc1ccc(Oc2ncccc2Br)cc1',\n",
       " 'CN1CC(=O)Nc2ccc(Br)cc21',\n",
       " 'O=C(OC[C@H]1O[C@@H](n2c3ccsc3c3c(Cl)ncnc32)[C@H](OC(=O)c2ccccc2)[C@@H]1OC(=O)c1ccccc1)c1ccccc1',\n",
       " 'Brc1ccc(N(c2ccccc2)c2ccccc2)cc1',\n",
       " 'COc1ccc(I)ccc1=O',\n",
       " 'Brc1oc2ccccc2c1Br',\n",
       " 'O=C1NCCc2cc(Br)ccc21',\n",
       " 'C/C=C(/c1ccccc1)n1ccc(=O)c2cc(I)ccc21',\n",
       " 'CCCCSc1nc(I)co1',\n",
       " 'O=C(OCc1ccccc1)c1cccc(Br)c1',\n",
       " 'COC(=O)c1c[nH]c2cc(Cl)c(Br)cc12',\n",
       " 'O=c1ccc2c(Cl)nc(NC(CO)CO)nc2n1-c1ccc(F)cc1F',\n",
       " 'O=C1C(I)=C[C@H]2CO[C@@H]1O2',\n",
       " 'O=c1[nH]c(=O)n(CCCN2C[C@@H]3C[C@]3(c3ccc(C(F)(F)F)cc3)C2)cc1I',\n",
       " 'CC(C)(C)[Si](C)(C)O[C@@H]1CC[C@]2(C)O[C@H]1[C@@H](CCO[Si](c1ccccc1)(c1ccccc1)C(C)(C)C)[C@H]2/C=C/I',\n",
       " 'Cc1cc(C=O)cc(C)c1Br',\n",
       " 'CC(C)(C)OC(=O)Nc1nc(Br)ns1',\n",
       " 'FC1(F)C(Cl)=C(Cl)C(F)(F)C1(F)F',\n",
       " 'CCOC(=O)CCc1c(C(=O)OCC)[nH]c2c([N+](=O)[O-])ccc(Cl)c12',\n",
       " 'Cc1nc2c(Br)c3oc(C)nc3c(Br)c2o1',\n",
       " 'C[C@H]1CCC[C@@H]2CC(=O)C(I)=C[C@]12C',\n",
       " 'COC(=O)c1cccc(Br)c1',\n",
       " 'Cn1ccc2c(Cl)nc3ccccc3c21',\n",
       " 'CCOC(=O)C1CCCN(CC/C=C/c2ccccc2Br)C1',\n",
       " 'CCN(CC)C(=O)OC(I)=C(F)F',\n",
       " 'O=S(=O)(OC1=CCCc2ccccc21)C(F)(F)F',\n",
       " 'CC(C)(C)OC(=O)N1CCC(O)(Cn2cnc(Cl)cc2=O)CC1',\n",
       " 'Brc1ccc2nc(C3CC(N4CCCCC4)C3)sc2c1',\n",
       " 'Clc1cc(Cl)nc(Cl)c1',\n",
       " 'CCCC[Te]c1ccccc1C',\n",
       " 'COc1nc(OC)nc(Oc2ccc(C3(c4ccc(Oc5nc(OC)nc(OC)n5)cc4)OC(=O)c4ccccc43)cc2)n1',\n",
       " 'CCNc1cccc(Br)n1',\n",
       " 'CC(=O)c1ccc(I)cc1',\n",
       " 'Brc1cn2ccnc2c(-c2ccc3ncn(C4CCCCC4)c3c2)n1',\n",
       " 'O=C1CCC2(CCCCC2)c2sc(Cl)cc21',\n",
       " 'Clc1ccc2c(ccc3ccc4ccc5cc(Cl)ccc5c4c32)c1',\n",
       " 'O=c1oc(Cl)cc2ccccc12',\n",
       " 'Oc1c(Br)cc2ccccc2c1-c1c(O)c(Br)cc2ccccc12',\n",
       " 'Ic1cccc(I)c1',\n",
       " 'FC(F)(F)CI',\n",
       " 'N#CCc1ccc(Br)cc1',\n",
       " 'Brc1ccc(-n2c(-c3ccccc3)ccc2-c2ccccc2)cc1',\n",
       " 'O=C(C=Cc1ccccc1)c1ccccc1',\n",
       " 'CC(C)(C)OC(=O)Oc1ccc(N=Nc2c(NC(=O)c3ccc(Br)cc3)nn(C(=O)OC(C)(C)C)c2N)cc1',\n",
       " 'CC(C)[Si](C#Cc1ccc(Br)cc1)(C(C)C)C(C)C',\n",
       " 'O=[N+]([O-])c1ccccc1I',\n",
       " 'COc1c(C(C)(C)C)cc(-c2cc(C(C)(C)C)c(OC)c(Br)c2C)c(C)c1Br',\n",
       " 'O=Cc1cccc(Br)n1',\n",
       " 'FC(Br)=C1O[C@H](COCc2ccccc2)[C@@H](OCc2ccccc2)[C@H](OCc2ccccc2)[C@H]1OCc1ccccc1',\n",
       " 'Brc1ccc(-c2ccccc2)cc1',\n",
       " 'COc1cc(Br)c(CCNC(=O)OCc2ccccc2)cc1OC',\n",
       " 'Nc1nccc2cc(Cl)oc12',\n",
       " 'COCC1=C(I)C(=O)CCC1',\n",
       " 'Clc1cccnc1',\n",
       " 'O=C(Nc1cccc(F)c1)c1cc(=O)[nH]c2cc(Br)ccc12',\n",
       " 'COc1cc(OC)cc(C2=Cc3ccccc3C(CC(=O)c3ccco3)N2c2ccc(I)cc2)c1',\n",
       " 'OC1CC(c2nc3ccc(Br)cc3s2)C1',\n",
       " 'Clc1ccc(Br)c2[nH]ccc12',\n",
       " 'N#Cc1cccc(Br)n1',\n",
       " 'O=[N+]([O-])c1ccccc1-c1cn2nc(Br)sc2n1',\n",
       " 'CC(C)(C)[Si](C)(C)OC[C@H]1O[C@@H](c2cccc(Br)n2)[C@H](O[Si](C)(C)C(C)(C)C)[C@@H]1O[Si](C)(C)C(C)(C)C',\n",
       " 'CCc1nc2ccc(I)cn2c1N(C)c1nc(-c2ccc(F)c(F)c2)cs1',\n",
       " 'CCCCCCC1(CCCCCC)c2cc(Br)ccc2-c2ccc(Br)cc21',\n",
       " 'O=S(=O)(Oc1c(Br)cc(Br)c2cccnc12)C(F)(F)F',\n",
       " 'C[C@H]1CC[C@@H](Nc2ncc(Br)cn2)CC1',\n",
       " 'Brc1ccc(Br)c2ccccc12',\n",
       " 'COc1ccc(-n2cccc(Br)c2=O)cc1',\n",
       " 'Fc1c(F)c(F)c(-c2c(F)c(F)c(F)c(F)c2Br)c(Br)c1F',\n",
       " 'COc1cc(O)c(C=O)cc1Br',\n",
       " 'Clc1ccccn1',\n",
       " 'COC(=O)c1ccc2c(c1)c(I)cn2C(=O)OC(C)(C)C',\n",
       " 'O=C(Oc1ccccc1)c1cc(OS(=O)(=O)C(F)(F)F)c2ccccc2c1OS(=O)(=O)C(F)(F)F',\n",
       " 'BrC(Br)=Cc1ccccc1-c1csc2ccccc12',\n",
       " 'Cc1nc(N)c2c(I)cn(COCC[Si](C)(C)C)c2n1',\n",
       " 'C[C@@H](Oc1cc(-n2cnc3cc(Br)ccc32)sc1C(N)=O)c1ccccc1Cl',\n",
       " 'OC/C=C/c1ccccc1',\n",
       " 'COc1cc2c(cc1Br)CCC(=O)N2CC(=O)OC(C)(C)C',\n",
       " 'O=S(=O)(Oc1c(Cl)c(Cl)c(OS(=O)(=O)C(F)(F)F)c2ccccc12)C(F)(F)F',\n",
       " 'O=C1COc2c(I)cc3ccccc3c2-c2c(c(I)cc3ccccc23)OCC(=O)Nc2cccc(n2)NC(=O)c2cc(cc([N+](=O)[O-])c2)C(=O)Nc2cccc(n2)N1',\n",
       " 'COc1ccc(CN2C(=O)Cn3c2nc(Br)c3C)cc1',\n",
       " 'O=[N+]([O-])c1ccc(Cl)cc1',\n",
       " 'Nc1nc(-c2ccc(Br)cc2)cs1',\n",
       " 'COc1ccc(-c2ccc(-c3ccc(-c4ccc(-c5ccc(-c6ccc(-c7ccc(I)c(OCC(=O)OC(C)(C)C)c7)cc6OCC(=O)OC(C)(C)C)c(OCC(=O)OC(C)(C)C)c5)cc4OCC(=O)OC(C)(C)C)c(OCC(=O)OC(C)(C)C)c3)cc2OCC(=O)OC(C)(C)C)c(OCC(=O)OC(C)(C)C)c1',\n",
       " 'NC(=O)c1cc(-c2ccc(Cl)cc2)cc(Br)c1N',\n",
       " 'O=S(=O)(c1ccccc1)n1ccc2nc(OS(=O)(=O)C(F)(F)F)ccc21',\n",
       " 'Cc1cccc(Br)c1',\n",
       " 'COc1c(Br)cc(NC(=O)OC(C)(C)C)cc1C(C)(C)C',\n",
       " 'O=C(NCc1nnc2ccc(Br)cn12)c1ccccc1',\n",
       " 'CC(C)(C)OC(=O)N[C@@H](Cc1ccc(OCc2ccccc2)c(I)c1)C(=O)OCc1ccccc1',\n",
       " 'Brc1ccc(N(c2ccccc2)c2ccc(Br)cc2)cc1',\n",
       " 'Brc1cc(Br)c(C2OCCO2)s1',\n",
       " 'CCCCC=CI',\n",
       " 'Oc1ccc(F)cc1Br',\n",
       " 'Nc1ccc(Cl)nn1',\n",
       " 'COC(=O)c1ccc2c(c1)nnn2S(=O)(=O)C(F)(F)F',\n",
       " 'CCNC(=O)Nc1cccc(Br)n1',\n",
       " 'Cc1nc(-c2ccc(C)c(Br)c2)no1',\n",
       " 'COc1ccc(Cn2cc(Cn3cnc4ncc(Br)cc4c3=O)nn2)cc1',\n",
       " 'CCC1OC(=O)/C1=C(\\\\Cl)c1ccccc1',\n",
       " 'Brc1ccc(C=Cc2ncccn2)cc1',\n",
       " 'CS(=O)(=O)NNc1ccccc1',\n",
       " 'Cn1c(=O)c(Br)c(OS(=O)(=O)C(F)(F)F)c2ccccc21',\n",
       " 'Cc1cc2c(cc1Br)Cc1ccc(C(F)(F)F)cc1-2',\n",
       " 'CS(=O)(=O)c1ccc(Oc2cc(Br)cc([N+](=O)[O-])c2)cc1',\n",
       " 'CCCCCCc1c(-c2ccc(Br)s2)sc(-c2ccc(Br)s2)c1CCCCCC',\n",
       " 'CC(=O)c1ccc(Br)cc1',\n",
       " 'NC(=O)c1ccc(Cl)cc1',\n",
       " 'CCn1cc(-c2nc3ccccc3[nH]2)c2cc(Br)ccc21',\n",
       " 'Cc1ccc2cc(C(C)(C)C)cn2c1I',\n",
       " 'CCCCCn1cc(Br)nc1[N+](=O)[O-]',\n",
       " 'OCc1cc(C(F)(F)F)ccc1Br',\n",
       " 'Cc1ncccc1OS(C)(=O)=O',\n",
       " 'CCCc1ccc(-c2c(F)cc(-c3ccc(I)c(F)c3F)cc2F)cc1F',\n",
       " 'Br/C=C/c1ccccc1',\n",
       " 'CC(=O)Nc1cc(Cl)cc(Cl)n1',\n",
       " 'O=C1OCC(Cl)=C1Cl',\n",
       " 'BrC=Cc1ccccc1',\n",
       " 'O=P(O)(O)[C@@H](O)c1ccc(Br)cc1',\n",
       " 'COc1c(OS(=O)(=O)C(F)(F)F)cc2ccccc2c1-c1c(OC)c(OC)c(-c2c(OC)c(OS(=O)(=O)C(F)(F)F)cc3ccccc23)c2ccccc12',\n",
       " 'CCCCCCCCN1C(=O)c2ccc3c4c(Br)cc5c6c(ccc(c7c(Br)cc(c2c37)C1=O)c64)C(=O)N(CCCCCCCC)C5=O',\n",
       " 'Brc1ccc(-c2nc(-c3ccccc3)c(-c3ccccc3)o2)nc1',\n",
       " 'Cn1cc(Br)c2ccccc21',\n",
       " 'Clc1cccc(Cl)n1',\n",
       " 'FC(F)(F)c1ccc(CBr)cc1',\n",
       " 'Brc1cnn2cc(-c3ccc(OCCN4CCCCC4)cc3)cnc12',\n",
       " 'CCCCCCC1(CCCCCC)c2ccccc2-c2ccc(-c3ccc(-c4ccc(-c5ccc(-c6ccc(Br)s6)s5)s4)s3)cc21',\n",
       " 'C=C(Br)C[C@@H](C)[C@H]1OC(c2ccc(OC)cc2)O[C@@H]1[C@@H]1CC[C@H](CCCO[Si](c2ccccc2)(c2ccccc2)C(C)(C)C)O1',\n",
       " 'COc1ccc(CCc2ccc(I)c(Br)c2)cc1-c1cc(C2OCCCO2)ccc1OC',\n",
       " 'COc1ccccc1Cl',\n",
       " 'C[C@H](Oc1cc2sc(Br)nc2c(Cl)c1F)[C@@H](C)O',\n",
       " 'O=C1CCCC(=O)N1C(=O)c1ccc(C(F)(F)F)cc1',\n",
       " 'C=C(CBr)C(=O)OCC',\n",
       " 'O=C1CCC(c2ccccc2Br)=CN1Cc1ccccc1',\n",
       " 'Nc1c(C=O)c(Cl)nc2ccccc12',\n",
       " 'CN(C(=O)C#CI)c1ccccc1',\n",
       " 'COC(=O)c1cc(Cl)c2nc(NCc3ccccc3)nc(N(C)C)c2n1',\n",
       " 'CC(=O)N(C/C=C/Br)CCc1c[nH]c2ccccc12',\n",
       " 'FC(F)(F)c1ccccc1I',\n",
       " 'CN(C)S(=O)(=O)Oc1ccc2ncccc2c1',\n",
       " 'COC(=O)[C@@H]1C[C@H](OC(C)=O)C(=O)[C@H]2[C@@]1(C)CC[C@H]1C(=O)O[C@H](c3ccoc3Br)C[C@]21C',\n",
       " 'O=C1CCCc2ccc(Br)cc21',\n",
       " 'Brc1ccc2cncn2c1',\n",
       " 'Cc1c(C)c2c(c(C)c1Br)NC(=O)CO2',\n",
       " 'Ic1ccc(N(c2ccc(I)cc2)c2ccc(I)cc2)cc1',\n",
       " 'C[C@@H](c1ccc(F)cc1)n1nnc2cnc3ccc(Br)cc3c21',\n",
       " 'CCCC=C(Br)CO[Si](C(C)C)(C(C)C)C(C)C',\n",
       " 'COc1ccc(-c2nc(Cl)nc(C)c2C(C)=O)cc1',\n",
       " 'CC1C(=O)NN=C2COc3cc(Br)c(NC4CN(C(=O)OC(C)(C)C)C4)cc3N21',\n",
       " 'CC1(C)C(=CC=C2CCCC(C=CC3=[N+](CCC[N+](C)(C)C)c4ccc(S(=O)(=O)[O-])cc4C3(C)C)=C2Cl)N(CCC[N+](C)(C)C)c2ccc(S(=O)(=O)[O-])cc21',\n",
       " 'Clc1cc(Cl)c(Cl)nc1Cl',\n",
       " 'COc1ccccc1-c1ccc2c(OS(=O)(=O)C(F)(F)F)cccc2c1',\n",
       " 'Brc1ccc(-c2cccnc2)nc1',\n",
       " 'O=S(=O)(Oc1ccc(Cl)cc1)n1ccnc1',\n",
       " 'O=C(/C=C/c1ccc(Br)cc1)c1ccc2c(c1)OCO2',\n",
       " 'Nc1cccc(Br)n1',\n",
       " 'COC(=O)c1ccccc1Br',\n",
       " 'COC(=O)/C=C/CBr',\n",
       " 'COc1ccc(N(C)S(=O)(=O)c2ccc(Br)cc2)cc1',\n",
       " 'Brc1c2ccccc2c(-c2ccc(-c3nc4ccccc4o3)cc2)c2ccccc12',\n",
       " 'O=c1cc(N2CCOCC2)oc2c(OS(=O)(=O)C(F)(F)F)cccc12',\n",
       " 'CCc1nn2c(Br)cccc2c1N(CC1CCOCC1)CC1CC1',\n",
       " 'Clc1ccc(Br)s1',\n",
       " 'CCOC(=O)c1ccsc1Br',\n",
       " 'Cc1ccc(Oc2nc(Oc3ccc(C)cc3)nc(Oc3ccc(C)cc3)n2)cc1',\n",
       " 'Fc1cccc(C(F)(F)F)c1Br',\n",
       " 'CCNC(=O)c1c(C)sc2nc(CCl)cc(=O)n12',\n",
       " 'Fc1cccc2c1cc1n2COc2ccc(Cl)nc2-1',\n",
       " 'COc1c2nc(cc3cc(-c4cc(C(=O)OC(C)(C)C)cc(C(=O)OC(C)(C)C)c4)c([nH]3)c(Br)c3nc(cc4cc(-c5cc(C(=O)OC(C)(C)C)cc(C(=O)OC(C)(C)C)c5)c1[nH]4)C(C)(C)C3)C(C)(C)C2',\n",
       " 'CN(C)/C=N/c1c(C#N)sc2ncc(Br)cc12',\n",
       " 'COc1c(-c2ccccc2Br)cccc1C(C)(C)C',\n",
       " 'C#CC1(O)CCCCC1',\n",
       " 'COC(=O)Cn1c(=O)c(Cl)nc2ccc(Cl)cc21',\n",
       " 'Cc1ccc(S(=O)(=O)Oc2ccc(F)cc2)cc1',\n",
       " 'O=S(=O)(Oc1cccc(C(F)(F)F)c1)C(F)(F)F',\n",
       " 'CSc1ccc(C(=O)c2ccc(Br)c(Cl)c2)cc1',\n",
       " 'Clc1ccc(Br)c(Cl)c1',\n",
       " 'Cc1nc2cc(Nc3nc(Cl)nc(Nc4ccc(F)cc4)n3)ccc2n1Cc1ccccc1',\n",
       " 'COc1ccc(NN)cc1',\n",
       " 'COc1ccc2c(c1)C(Br)=CC2=O',\n",
       " 'O=CC1=C(Br)c2ccccc2CC1',\n",
       " 'CC(C)(C)OC(=O)NCCOc1cccc2nc(Cl)sc12',\n",
       " 'O=C1C=Cc2cc(Br)cc3ccc(O)c1c23',\n",
       " 'Nc1ncnc2ccc(Br)cc12',\n",
       " 'Brc1cccc(-c2ccc3c(c2)c2ccccc2n3-c2ccccc2)c1',\n",
       " 'COc1nc(Br)cnc1I',\n",
       " 'C=CCON1C(=O)N2C[C@H]1C(I)=C[C@H]2CO[Si](C)(C)C(C)(C)C',\n",
       " 'OCc1ccnc(Br)c1',\n",
       " 'CN(C)C(=O)N[C@@H]1CCN(c2n[nH]c3nccc(Cl)c23)C1',\n",
       " 'COc1ccc(CNc2nc(Cl)nc3c2ncn3C(C)C)cc1',\n",
       " 'CCC(C)(OC(C)=O)c1ccc2ccccc2c1',\n",
       " 'COC(=O)N1CCC2(CC1)C(=O)N(Cc1ncc3cc(Cl)ccc3c1Br)c1cnccc12',\n",
       " 'Cc1ccc(OC/C=C/c2ccccc2)c(C#CBr)c1',\n",
       " 'Nc1ccc2[nH]nc(Br)c2c1',\n",
       " 'COCOc1cccc2occ(I)c(=O)c12',\n",
       " 'CCN(CC)N=Nc1cc(Br)cc(Br)c1',\n",
       " 'Cc1cc(C)cc(OS(=O)(=O)C(F)(F)F)c1',\n",
       " 'Brc1ccc2c(c1)C1(c3cc(Br)ccc3-2)c2cc(Br)ccc2-c2ccc(Br)cc21',\n",
       " 'O=Cc1cc(Br)ccc1O',\n",
       " 'CC(C)(C)S(=O)N[C@@H]1CCCc2ccc(Br)cc21',\n",
       " 'COc1ccc(-c2nc(OS(=O)(=O)C(F)(F)F)co2)cc1',\n",
       " 'CCN(CC)C(=O)c1c(OC)ccc2cc(Br)ccc12',\n",
       " 'CCCCCCCCCCCCCCCCc1cc(Br)cc(C#C[Si](C(C)C)(C(C)C)C(C)C)c1',\n",
       " 'Cc1nc(N)ccc1Br',\n",
       " 'Nc1ccccc1I',\n",
       " 'Clc1nc(-c2ccccc2)nc(-c2ccccc2)n1',\n",
       " 'Cc1ccc(S(=O)(=O)OCC2CCc3cc(F)cc(Br)c3O2)cc1',\n",
       " 'CC1CC(=O)C=C(I)N1C(=O)OC(C)(C)C',\n",
       " 'Brc1ccc(-c2cnco2)cc1',\n",
       " 'O=[N+]([O-])c1ccc(I)cc1',\n",
       " 'N#Cc1ccc(Cl)nc1',\n",
       " 'N#CCCCCCCBr',\n",
       " 'CCCc1cc(=O)oc2cc(I)c3c(c12)OC(C)(C)C=C3',\n",
       " 'CCOC(=O)c1[nH]c2ccc(Br)cc2c1I',\n",
       " 'CC(C)(C)OC(=O)N1CC(I)C1',\n",
       " 'CC(C)(C)c1ccc(-c2cc([N+](=O)[O-])cc(Br)c2N)cc1',\n",
       " 'Ic1cccc2cc[nH]c12',\n",
       " 'O=S(=O)(Oc1ccc(Cl)cc1)C(F)(F)F',\n",
       " 'N#CC(=C(c1ccccc1)c1ccccc1)c1ccc(Br)cc1',\n",
       " 'COc1ccc(C(=C(Br)C#N)c2ccc(OC)cc2)cc1',\n",
       " 'Brc1cc2c3c(c1)CCCN3CCC2',\n",
       " 'O=S(=O)(c1ccccc1)c1ccc(Cl)cc1',\n",
       " 'Cc1cccc(I)c1',\n",
       " 'Cc1ccc(S(=O)(=O)n2cc(-c3ccc(C#N)cc3)c3cc(Br)cnc32)cc1',\n",
       " 'Clc1nc(-c2ccccc2)nc(-c2ccc(-c3ccccc3)cc2)n1',\n",
       " 'Clc1c(C#Cc2ccccc2)c(-c2ccccc2)nc2ccccc12',\n",
       " 'C=Cc1ccc(Cl)cc1-c1ccccc1CC(C)=O',\n",
       " 'COc1cccc(Cl)n1',\n",
       " 'COc1ccc(-c2nnc(-c3ccc(I)cc3)o2)cc1',\n",
       " 'Nc1cc(Cl)nc(N)n1',\n",
       " 'Fc1c(F)c(F)c(C2=NCCO2)c(F)c1F',\n",
       " 'CC1=CC(=O)N(c2cc(C(F)(F)F)ccc2Br)C1=O',\n",
       " 'CC1CN(C)CC=C1OS(=O)(=O)C(F)(F)F',\n",
       " 'CC(C)(C)OC(=O)n1nc(I)c2ccccc21',\n",
       " 'Nc1ccc(OS(=O)(=O)F)cc1',\n",
       " 'COc1cccc(CCNC(=O)Cc2ccc(OC)c(OC)c2Br)c1OC',\n",
       " 'COc1cc(N(C)c2ccc(N(C)C)c(C#C/C=C/Cl)c2)cc(OC)c1OC',\n",
       " 'Clc1ccc2ccccc2n1',\n",
       " 'Oc1c(Br)cc(Br)cc1Br',\n",
       " 'CCOC(=O)CN(C(=O)C1(C)CC1)C1CCC(Oc2cc(=O)n(C)cc2Br)CC1',\n",
       " 'Nc1ccc(Br)cc1[N+](=O)[O-]',\n",
       " 'Fc1cc(Br)ccc1-c1ccccc1',\n",
       " 'Cl/C=C/c1sc2ccccc2c1Br',\n",
       " 'COc1ccc2ccccc2c1-c1c(OC)ccc2cc(-c3cccc(-c4cccc(Br)n4)n3)ccc12',\n",
       " 'COC(=O)[C@H](Cc1cccc(OS(=O)(=O)C(F)(F)F)c1)NC(=O)OC(C)(C)C',\n",
       " 'OCc1cccc2c(Br)cccc12',\n",
       " 'Cc1cc2oc(=O)cc(-c3ccccc3)c2c(C)c1Br',\n",
       " 'CC(=O)Nc1ccc(C(F)(F)F)cc1Br',\n",
       " 'CN(C)c1ccc(C=CC=C2C(=O)c3c(I)ccc(I)c3C2=O)cc1',\n",
       " 'ClC1=C(Br)CCCC1',\n",
       " 'FC(F)(F)c1ccc(I)cc1',\n",
       " 'O=C1c2ccccc2-c2c(-c3ccc(Br)cc3)ccc(-c3ccc(Br)cc3)c21',\n",
       " 'Clc1ncccc1Br',\n",
       " 'COC(=O)/C(=C/Br)NC(=O)[C@@H](Cc1ccccc1)NC(=O)OC(C)(C)C',\n",
       " 'O=c1c(Br)cn(-c2ccccc2)c(=O)n1CC1CC1',\n",
       " 'Cn1cc(-c2ccccc2I)c2ccccc21',\n",
       " 'O=C(Oc1ccccc1)c1cc2ccccc2o1',\n",
       " 'CC(C)(O)c1cc(I)c(-c2ccccc2)[se]1',\n",
       " 'CC1CCCC=C1OS(=O)(=O)C(F)(F)F',\n",
       " 'COc1nn2c(Br)cccc2c1N(CC1CCOCC1)CC1CC1',\n",
       " 'FC1(F)c2cccc(c2)C(F)(F)C(F)(F)c2cc(ccc2I)C1(F)F',\n",
       " 'Clc1ccccc1',\n",
       " 'FC(F)(F)c1ccc(Br)nc1',\n",
       " 'Clc1cncc(Cl)n1',\n",
       " 'CCCc1ccc(-c2c(F)cc(-c3ccc(I)c(F)c3)cc2F)cc1',\n",
       " 'CS(=O)(=O)Oc1cccc2ccccc12',\n",
       " 'IC1COC1',\n",
       " 'CC1(C)C=Cc2c(ccc3c(=O)c(I)coc23)O1',\n",
       " 'COC(=O)c1scc(Br)c1OS(=O)(=O)C(F)(F)F',\n",
       " 'CCCCNc1nc(N)c2c(Br)nn(C(C)(C)C)c2n1',\n",
       " 'COc1ccc(Br)cc1',\n",
       " 'Cc1ccc(NNS(C)(=O)=O)cc1',\n",
       " 'CCCCCCCCCCCCOc1c2cc(Br)sc2c(OCCCCCCCCCCCC)c2cc(Br)sc12',\n",
       " 'COc1ccc(N2c3ccc(Br)cc3Sc3cc(Br)ccc32)cc1',\n",
       " 'CCOC(=O)[C@@H]1C2c3ccccc3C(c3ccccc32)[C@H]1c1cccc(Cl)n1',\n",
       " 'COc1cc2c(c(OC)c1OC)CC[C@H](O[Si](c1ccccc1)(c1ccccc1)C(C)(C)C)C=C2OS(=O)(=O)C(F)(F)F',\n",
       " 'Brc1ccccc1',\n",
       " 'Cc1ccc(Br)nc1',\n",
       " 'CCCCCCN1c2ccc(Br)cc2Sc2cc(C=O)ccc21',\n",
       " 'O=C(O)Cc1cc(-c2ccc(Br)s2)sc1-c1ccc(-c2sc(-c3ccc(Br)s3)cc2CC(=O)O)s1',\n",
       " 'CC(C)(C)OC(=O)N1CCC(OC(=O)N2CCc3ncc(Br)cc32)CC1',\n",
       " 'COc1ccc2nc(Br)sc2c1',\n",
       " 'OCc1cccc2c(-c3ccc(Br)cc3)cccc12',\n",
       " 'CCOC(=O)c1c(I)c2cc(OCc3ccccc3)ccc2n1S(=O)(=O)c1ccccc1',\n",
       " 'Nc1ncc(Br)nc1Nc1ccc2[nH]ccc2c1',\n",
       " 'CC(C)(O)C(=O)Cc1ccc(Br)cc1',\n",
       " 'O=C1CC(c2ccccc2)C2(C(=O)c3ccccc3C2=O)C(c2ccc(Br)cc2)C1',\n",
       " 'Cc1[nH]c2c(C(N)=O)ccc(Br)c2c1C',\n",
       " 'O=C1c2cccc(I)c2-n2cccc21',\n",
       " 'Ic1cnc(-c2ccccc2)s1',\n",
       " 'COC(=O)Cc1ccsc1-c1ccc(Cl)s1',\n",
       " 'O=S(=O)(n1nnc2ccccc21)C(F)(F)F',\n",
       " 'OCCOc1ccc2cc(Br)ccc2c1-c1c(OCCO)ccc2cc(Br)ccc12',\n",
       " 'Brc1ccc(-c2nc3c4ccccc4ccc3n2-c2ccccc2)cc1',\n",
       " 'Clc1cccc(CBr)c1',\n",
       " 'O=c1cc(Cl)c2cccnc2n1Cc1ccccc1',\n",
       " 'O=Cc1cccnc1Br',\n",
       " 'CC1(C)C(I)=C(I)c2c1ccc1ccccc21',\n",
       " 'Brc1ccc2ccccc2c1',\n",
       " 'O=Cc1ccnc(Br)c1',\n",
       " 'CCCc1c(F)cc(-c2ccc(-c3ccc(I)c(F)c3)cc2)cc1F',\n",
       " 'Cc1cc(Br)cc2nnc(N)nc12',\n",
       " 'C=CCc1cccc(C=O)c1OS(=O)(=O)C(F)(F)F',\n",
       " 'Brc1cc(CCOC2CCCCO2)c(-c2ccncc2)s1',\n",
       " 'O=Cc1ccc(-n2c3ccc(Br)cc3c3cc(Br)ccc32)cc1',\n",
       " 'O=C1CCC(N2C(=O)c3ccc(Br)cc3C2=O)C(=O)N1',\n",
       " 'Fc1ccc(-c2cn3nc(Cl)ccc3n2)cc1',\n",
       " 'Clc1ccc2nccn2n1',\n",
       " 'CCCCCCCCCCCCc1ccc(-c2cc(I)c(-c3ccc(CCCCCCCCCCCC)cc3)cc2I)cc1',\n",
       " 'COC(=O)[C@@H]1CN1S(=O)(=O)c1ccc(C)cc1',\n",
       " 'O=c1cc(Cl)nc2ccccn12',\n",
       " 'COc1cc(C)cc(-c2nc(Nc3cc(C(F)(F)F)cc(C(F)(F)F)c3)ncc2-c2ccnc(Cl)c2)c1',\n",
       " 'CCCn1nc(Br)c2nnn(Cc3ccc(Cl)cc3)c(=O)c21',\n",
       " 'COC(=O)c1nc(Cl)ccc1Cl',\n",
       " 'O=Cc1c(Cl)ccnc1N1CCn2c(cc3c2CCCC3)C1=O',\n",
       " 'Nc1cc(Cl)nn2ccnc12',\n",
       " 'CCOC(=O)c1cc2c(=O)c3cc(Br)c(F)cc3oc2nc1N',\n",
       " 'COC(=O)c1cc(OC)c2c(Br)ccnc2c1OC',\n",
       " 'Cc1ccc(C[N+](C)(C)C)cc1',\n",
       " 'CCCc1ccc(-c2c(F)cc(-c3ccc(I)cc3)cc2F)cc1F',\n",
       " 'Brc1ccc2c(c1)n1c3ccccc3nc1n2-c1ccccc1',\n",
       " 'CCCCC(CC)CN1C(=O)C2=C(c3ccc(Br)s3)N(CC(CC)CCCC)C(=O)C2=C1c1ccc(Br)s1',\n",
       " 'COc1ccc2ccc(OS(=O)(=O)C(F)(F)F)c(Cl)c2c1',\n",
       " 'COC(=O)c1ccc(Br)o1',\n",
       " 'Nc1cccnc1Cl',\n",
       " 'O=S(=O)(NC1Cc2ccc(Br)cc2C1)c1ccccc1',\n",
       " 'Nc1ncnc2c1c(I)cn2[C@H]1C[C@H](O)[C@@H](CO)O1',\n",
       " 'O=Cc1ccc(OCc2cccc(Br)c2)cc1',\n",
       " 'Nc1c(F)cc(Br)cc1F',\n",
       " 'C=C(OS(=O)(=O)c1ccc(C)cc1)P(=O)(OCC)OCC',\n",
       " 'COc1ccc(C2=CN3C(=O)c4cc(OC)c(OCCCOc5cc6c(cc5OC)C(=O)N5C=C(OS(=O)(=O)C(F)(F)F)C[C@H]5C(=O)N6COCC[Si](C)(C)C)cc4N(COCC[Si](C)(C)C)C(=O)[C@@H]3C2)cc1',\n",
       " 'CNc1nc(C)c2cc(Br)cc(OCC3CCOCC3)c2n1',\n",
       " 'O/N=C(\\\\CC(c1ccccc1)c1ccc(Br)cc1)c1ccncc1',\n",
       " 'CCOC(=O)c1ccccc1Sc1ccc(Br)cc1',\n",
       " 'CCCc1c(F)cc(-c2c(F)cc(-c3ccc(I)cc3)cc2F)cc1F',\n",
       " 'COc1ccc(OC(=O)C(C)(C)C)cc1',\n",
       " 'CC(C)(C)OC(=O)N(Cc1ccccc1)C(=O)C1CCCCC1',\n",
       " 'O=C1OC(C2CCCCC2)/C1=C(/Cl)c1ccccc1',\n",
       " 'COC(=O)c1ccc(-c2c3nc(c(Br)c4ccc([nH]4)c(-c4ccc(C(=O)OC)cc4)c4nc(c(Br)c5ccc2[nH]5)C=C4)C=C3)cc1',\n",
       " 'O=S(=O)(Oc1coc(-c2ccccc2)n1)C(F)(F)F',\n",
       " 'O=c1[nH]c(=O)n([C@@H]2O[C@H](COP(=O)(O)OP(=O)(O)OP(=O)(O)O)[C@@H](O)[C@H]2O)cc1I',\n",
       " 'COC(=O)[C@H](Cc1ccccc1Br)NC(=O)OCc1ccccc1',\n",
       " 'Brc1cnc(-c2ccccc2)[nH]1',\n",
       " 'CC(C)Cn1nc2c(-c3ccc(Br)s3)nc3ncnn3c2n1',\n",
       " 'COc1ccc(-c2cccc(-c3ccc(OC)cc3)c2I)cc1',\n",
       " 'NCc1ccc(Br)cc1',\n",
       " 'Brc1cccc(-c2nc(-c3ccccc3)nc(-c3ccccc3)n2)c1',\n",
       " 'O=C(Nc1cccc(Br)c1)Nc1cccc(C(F)(F)F)c1',\n",
       " 'Clc1ccc(I)cc1Br',\n",
       " 'CCCCNC(=O)Nc1ccnc2cc(Br)cnc12',\n",
       " 'CC(C)=CCl',\n",
       " 'COc1ccc(C#Cc2nccnc2Cl)cc1',\n",
       " 'CC(C)(C)c1cc(Br)cc(C=O)c1O',\n",
       " 'CCCCCCCCc1cc(Br)sc1-c1nc2sc(-c3sc(C=O)cc3CCCCCCCC)nc2s1',\n",
       " 'Fc1nc(F)c(Cl)c(F)c1Cl',\n",
       " 'Cc1ccc(-c2nn(CCc3ccccc3)nc2Br)cc1',\n",
       " 'CCOC(=O)c1cc2c(Br)c(OC)c(OC)c(Br)c2[nH]1',\n",
       " 'COC(=O)C=CI',\n",
       " 'CC(C)[Si](Oc1ccc2c(c1)OC(C)(C)C(Br)=C2)(C(C)C)C(C)C',\n",
       " 'OC[C@@H](Nc1ncnc2cc(Br)sc12)c1ccccc1',\n",
       " 'O=C(c1ccccc1)c1c[nH]c2ccc(Br)cc12',\n",
       " 'Brc1c2ccccc2c(-c2cccs2)c2ccccc12',\n",
       " 'COC(=O)C1CN1S(=O)(=O)c1ccccc1[N+](=O)[O-]',\n",
       " 'Cc1ccc(S(=O)(=O)Oc2ccc3ncccc3c2)cc1',\n",
       " 'Ic1cccc(OC2CCC2)c1',\n",
       " 'CN(C)CCCNC(=O)c1cc(NC(=O)c2cc(NC(=O)c3ccc(Br)s3)cn2C)cn1C',\n",
       " 'O=C(c1ccccc1)c1ccc(Br)nc1',\n",
       " 'COC(OC)c1cc(N2CCN(C)CC2)nc(SC)n1',\n",
       " 'O=C1C(Cl)=C(c2ccccc2)N(c2ccccc2)/C1=C/c1cccs1',\n",
       " 'COc1c2nc(cc3cc(Br)c(cc4nc(cc5cc(Br)c1[nH]5)C(C)(C)C4)[nH]3)C(C)(C)C2',\n",
       " 'CCOC(=O)C(C)(C)Br',\n",
       " 'Ic1ccc(C2(c3ccc(N4c5ccccc5Oc5ccccc54)cc3)CCCCC2)cc1',\n",
       " 'O=C1OCC(Br)=C1Br',\n",
       " 'O=C1CCc2ccc(Br)cc21',\n",
       " 'COC(=O)c1ccccc1I',\n",
       " ...]"
      ]
     },
     "execution_count": 281,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "[canonicalize_smiles(smiles) for smiles in  data['Component_1'].unique()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 272,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 54297/54297 [00:00<00:00, 987229.82it/s]\n"
     ]
    }
   ],
   "source": [
    "chosen_reactions, test_molecules = split_train_test_reactions(reaction_components)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 284,
   "metadata": {},
   "outputs": [
    {
     "ename": "KeyError",
     "evalue": "'COc1ccc(-c2ncc(I)c(-c3ccccc3)c2C(=O)c2ccccc2)cc1'",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mKeyError\u001b[0m                                  Traceback (most recent call last)",
      "\u001b[1;32mf:\\programming_projects\\yield_prediction\\yield_prediction.ipynb Cell 49'\u001b[0m in \u001b[0;36m<cell line: 1>\u001b[1;34m()\u001b[0m\n\u001b[1;32m----> <a href='vscode-notebook-cell:/f%3A/programming_projects/yield_prediction/yield_prediction.ipynb#ch0000038?line=0'>1</a>\u001b[0m u_train, u_test \u001b[39m=\u001b[39m make_symbolic_reactions(data\u001b[39m.\u001b[39;49miloc[\u001b[39msorted\u001b[39;49m(chosen_reactions)], data\u001b[39m.\u001b[39;49miloc[\u001b[39msorted\u001b[39;49m(test_molecules)])\n",
      "\u001b[1;32mf:\\programming_projects\\yield_prediction\\yield_prediction.ipynb Cell 58'\u001b[0m in \u001b[0;36mmake_symbolic_reactions\u001b[1;34m(*reaction_dfs)\u001b[0m\n\u001b[0;32m      <a href='vscode-notebook-cell:/f%3A/programming_projects/yield_prediction/yield_prediction.ipynb#ch0000119?line=5'>6</a>\u001b[0m         unique_molecules\u001b[39m.\u001b[39mupdate(canonicalize_smiles(smiles) \u001b[39mfor\u001b[39;00m smiles \u001b[39min\u001b[39;00m reaction_df[col]\u001b[39m.\u001b[39munique())\n\u001b[0;32m      <a href='vscode-notebook-cell:/f%3A/programming_projects/yield_prediction/yield_prediction.ipynb#ch0000119?line=6'>7</a>\u001b[0m mol_id \u001b[39m=\u001b[39m {smiles: i \u001b[39mfor\u001b[39;00m smiles \u001b[39min\u001b[39;00m \u001b[39menumerate\u001b[39m(\u001b[39msorted\u001b[39m(unique_molecules))}\n\u001b[1;32m----> <a href='vscode-notebook-cell:/f%3A/programming_projects/yield_prediction/yield_prediction.ipynb#ch0000119?line=7'>8</a>\u001b[0m \u001b[39mreturn\u001b[39;00m [reaction_df[[\u001b[39m'\u001b[39;49m\u001b[39mComponent_1\u001b[39;49m\u001b[39m'\u001b[39;49m, \u001b[39m'\u001b[39;49m\u001b[39mComponent_2\u001b[39;49m\u001b[39m'\u001b[39;49m, \u001b[39m'\u001b[39;49m\u001b[39mProduct\u001b[39;49m\u001b[39m'\u001b[39;49m]]\u001b[39m.\u001b[39;49mapplymap(\u001b[39mlambda\u001b[39;49;00m x: mol_id[canonicalize_smiles(x)])]\n",
      "File \u001b[1;32mf:\\Anaconda\\envs\\yield-prediction\\lib\\site-packages\\pandas\\core\\frame.py:8924\u001b[0m, in \u001b[0;36mDataFrame.applymap\u001b[1;34m(self, func, na_action, **kwargs)\u001b[0m\n\u001b[0;32m   8921\u001b[0m         \u001b[39mreturn\u001b[39;00m lib\u001b[39m.\u001b[39mmap_infer(x, func, ignore_na\u001b[39m=\u001b[39mignore_na)\n\u001b[0;32m   8922\u001b[0m     \u001b[39mreturn\u001b[39;00m lib\u001b[39m.\u001b[39mmap_infer(x\u001b[39m.\u001b[39mastype(\u001b[39mobject\u001b[39m)\u001b[39m.\u001b[39m_values, func, ignore_na\u001b[39m=\u001b[39mignore_na)\n\u001b[1;32m-> 8924\u001b[0m \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39;49m\u001b[39m.\u001b[39;49mapply(infer)\u001b[39m.\u001b[39m__finalize__(\u001b[39mself\u001b[39m, \u001b[39m\"\u001b[39m\u001b[39mapplymap\u001b[39m\u001b[39m\"\u001b[39m)\n",
      "File \u001b[1;32mf:\\Anaconda\\envs\\yield-prediction\\lib\\site-packages\\pandas\\core\\frame.py:8839\u001b[0m, in \u001b[0;36mDataFrame.apply\u001b[1;34m(self, func, axis, raw, result_type, args, **kwargs)\u001b[0m\n\u001b[0;32m   8828\u001b[0m \u001b[39mfrom\u001b[39;00m \u001b[39mpandas\u001b[39;00m\u001b[39m.\u001b[39;00m\u001b[39mcore\u001b[39;00m\u001b[39m.\u001b[39;00m\u001b[39mapply\u001b[39;00m \u001b[39mimport\u001b[39;00m frame_apply\n\u001b[0;32m   8830\u001b[0m op \u001b[39m=\u001b[39m frame_apply(\n\u001b[0;32m   8831\u001b[0m     \u001b[39mself\u001b[39m,\n\u001b[0;32m   8832\u001b[0m     func\u001b[39m=\u001b[39mfunc,\n\u001b[1;32m   (...)\u001b[0m\n\u001b[0;32m   8837\u001b[0m     kwargs\u001b[39m=\u001b[39mkwargs,\n\u001b[0;32m   8838\u001b[0m )\n\u001b[1;32m-> 8839\u001b[0m \u001b[39mreturn\u001b[39;00m op\u001b[39m.\u001b[39;49mapply()\u001b[39m.\u001b[39m__finalize__(\u001b[39mself\u001b[39m, method\u001b[39m=\u001b[39m\u001b[39m\"\u001b[39m\u001b[39mapply\u001b[39m\u001b[39m\"\u001b[39m)\n",
      "File \u001b[1;32mf:\\Anaconda\\envs\\yield-prediction\\lib\\site-packages\\pandas\\core\\apply.py:727\u001b[0m, in \u001b[0;36mFrameApply.apply\u001b[1;34m(self)\u001b[0m\n\u001b[0;32m    724\u001b[0m \u001b[39melif\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39mraw:\n\u001b[0;32m    725\u001b[0m     \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39mapply_raw()\n\u001b[1;32m--> 727\u001b[0m \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39;49m\u001b[39m.\u001b[39;49mapply_standard()\n",
      "File \u001b[1;32mf:\\Anaconda\\envs\\yield-prediction\\lib\\site-packages\\pandas\\core\\apply.py:851\u001b[0m, in \u001b[0;36mFrameApply.apply_standard\u001b[1;34m(self)\u001b[0m\n\u001b[0;32m    850\u001b[0m \u001b[39mdef\u001b[39;00m \u001b[39mapply_standard\u001b[39m(\u001b[39mself\u001b[39m):\n\u001b[1;32m--> 851\u001b[0m     results, res_index \u001b[39m=\u001b[39m \u001b[39mself\u001b[39;49m\u001b[39m.\u001b[39;49mapply_series_generator()\n\u001b[0;32m    853\u001b[0m     \u001b[39m# wrap results\u001b[39;00m\n\u001b[0;32m    854\u001b[0m     \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39mwrap_results(results, res_index)\n",
      "File \u001b[1;32mf:\\Anaconda\\envs\\yield-prediction\\lib\\site-packages\\pandas\\core\\apply.py:867\u001b[0m, in \u001b[0;36mFrameApply.apply_series_generator\u001b[1;34m(self)\u001b[0m\n\u001b[0;32m    864\u001b[0m \u001b[39mwith\u001b[39;00m option_context(\u001b[39m\"\u001b[39m\u001b[39mmode.chained_assignment\u001b[39m\u001b[39m\"\u001b[39m, \u001b[39mNone\u001b[39;00m):\n\u001b[0;32m    865\u001b[0m     \u001b[39mfor\u001b[39;00m i, v \u001b[39min\u001b[39;00m \u001b[39menumerate\u001b[39m(series_gen):\n\u001b[0;32m    866\u001b[0m         \u001b[39m# ignore SettingWithCopy here in case the user mutates\u001b[39;00m\n\u001b[1;32m--> 867\u001b[0m         results[i] \u001b[39m=\u001b[39m \u001b[39mself\u001b[39;49m\u001b[39m.\u001b[39;49mf(v)\n\u001b[0;32m    868\u001b[0m         \u001b[39mif\u001b[39;00m \u001b[39misinstance\u001b[39m(results[i], ABCSeries):\n\u001b[0;32m    869\u001b[0m             \u001b[39m# If we have a view on v, we need to make a copy because\u001b[39;00m\n\u001b[0;32m    870\u001b[0m             \u001b[39m#  series_generator will swap out the underlying data\u001b[39;00m\n\u001b[0;32m    871\u001b[0m             results[i] \u001b[39m=\u001b[39m results[i]\u001b[39m.\u001b[39mcopy(deep\u001b[39m=\u001b[39m\u001b[39mFalse\u001b[39;00m)\n",
      "File \u001b[1;32mf:\\Anaconda\\envs\\yield-prediction\\lib\\site-packages\\pandas\\core\\frame.py:8922\u001b[0m, in \u001b[0;36mDataFrame.applymap.<locals>.infer\u001b[1;34m(x)\u001b[0m\n\u001b[0;32m   8920\u001b[0m \u001b[39mif\u001b[39;00m x\u001b[39m.\u001b[39mempty:\n\u001b[0;32m   8921\u001b[0m     \u001b[39mreturn\u001b[39;00m lib\u001b[39m.\u001b[39mmap_infer(x, func, ignore_na\u001b[39m=\u001b[39mignore_na)\n\u001b[1;32m-> 8922\u001b[0m \u001b[39mreturn\u001b[39;00m lib\u001b[39m.\u001b[39;49mmap_infer(x\u001b[39m.\u001b[39;49mastype(\u001b[39mobject\u001b[39;49m)\u001b[39m.\u001b[39;49m_values, func, ignore_na\u001b[39m=\u001b[39;49mignore_na)\n",
      "File \u001b[1;32mf:\\Anaconda\\envs\\yield-prediction\\lib\\site-packages\\pandas\\_libs\\lib.pyx:2870\u001b[0m, in \u001b[0;36mpandas._libs.lib.map_infer\u001b[1;34m()\u001b[0m\n",
      "\u001b[1;32mf:\\programming_projects\\yield_prediction\\yield_prediction.ipynb Cell 58'\u001b[0m in \u001b[0;36mmake_symbolic_reactions.<locals>.<lambda>\u001b[1;34m(x)\u001b[0m\n\u001b[0;32m      <a href='vscode-notebook-cell:/f%3A/programming_projects/yield_prediction/yield_prediction.ipynb#ch0000119?line=5'>6</a>\u001b[0m         unique_molecules\u001b[39m.\u001b[39mupdate(canonicalize_smiles(smiles) \u001b[39mfor\u001b[39;00m smiles \u001b[39min\u001b[39;00m reaction_df[col]\u001b[39m.\u001b[39munique())\n\u001b[0;32m      <a href='vscode-notebook-cell:/f%3A/programming_projects/yield_prediction/yield_prediction.ipynb#ch0000119?line=6'>7</a>\u001b[0m mol_id \u001b[39m=\u001b[39m {smiles: i \u001b[39mfor\u001b[39;00m smiles \u001b[39min\u001b[39;00m \u001b[39menumerate\u001b[39m(\u001b[39msorted\u001b[39m(unique_molecules))}\n\u001b[1;32m----> <a href='vscode-notebook-cell:/f%3A/programming_projects/yield_prediction/yield_prediction.ipynb#ch0000119?line=7'>8</a>\u001b[0m \u001b[39mreturn\u001b[39;00m [reaction_df[[\u001b[39m'\u001b[39m\u001b[39mComponent_1\u001b[39m\u001b[39m'\u001b[39m, \u001b[39m'\u001b[39m\u001b[39mComponent_2\u001b[39m\u001b[39m'\u001b[39m, \u001b[39m'\u001b[39m\u001b[39mProduct\u001b[39m\u001b[39m'\u001b[39m]]\u001b[39m.\u001b[39mapplymap(\u001b[39mlambda\u001b[39;00m x: mol_id[canonicalize_smiles(x)])]\n",
      "\u001b[1;31mKeyError\u001b[0m: 'COc1ccc(-c2ncc(I)c(-c3ccccc3)c2C(=O)c2ccccc2)cc1'"
     ]
    }
   ],
   "source": [
    "u_train, u_test = make_symbolic_reactions(data.iloc[sorted(chosen_reactions)], data.iloc[sorted(test_molecules)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 273,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[4,\n",
       " 9,\n",
       " 23,\n",
       " 30,\n",
       " 32,\n",
       " 39,\n",
       " 40,\n",
       " 41,\n",
       " 43,\n",
       " 55,\n",
       " 66,\n",
       " 72,\n",
       " 81,\n",
       " 94,\n",
       " 113,\n",
       " 119,\n",
       " 120,\n",
       " 141,\n",
       " 146,\n",
       " 156,\n",
       " 165,\n",
       " 185,\n",
       " 188,\n",
       " 196,\n",
       " 198,\n",
       " 219,\n",
       " 224,\n",
       " 230,\n",
       " 244,\n",
       " 266,\n",
       " 289,\n",
       " 302,\n",
       " 313,\n",
       " 322,\n",
       " 334,\n",
       " 342,\n",
       " 344,\n",
       " 360,\n",
       " 367,\n",
       " 369,\n",
       " 385,\n",
       " 393,\n",
       " 419,\n",
       " 431,\n",
       " 442,\n",
       " 455,\n",
       " 466,\n",
       " 498,\n",
       " 500,\n",
       " 502,\n",
       " 505,\n",
       " 508,\n",
       " 534,\n",
       " 549,\n",
       " 560,\n",
       " 570,\n",
       " 608,\n",
       " 629,\n",
       " 639,\n",
       " 641,\n",
       " 642,\n",
       " 658,\n",
       " 665,\n",
       " 670,\n",
       " 687,\n",
       " 696,\n",
       " 704,\n",
       " 711,\n",
       " 722,\n",
       " 754,\n",
       " 755,\n",
       " 764,\n",
       " 769,\n",
       " 787,\n",
       " 794,\n",
       " 797,\n",
       " 803,\n",
       " 805,\n",
       " 834,\n",
       " 842,\n",
       " 853,\n",
       " 874,\n",
       " 889,\n",
       " 918,\n",
       " 924,\n",
       " 930,\n",
       " 982,\n",
       " 1010,\n",
       " 1022,\n",
       " 1040,\n",
       " 1042,\n",
       " 1044,\n",
       " 1055,\n",
       " 1065,\n",
       " 1076,\n",
       " 1080,\n",
       " 1088,\n",
       " 1105,\n",
       " 1125,\n",
       " 1129,\n",
       " 1133,\n",
       " 1152,\n",
       " 1164,\n",
       " 1172,\n",
       " 1173,\n",
       " 1181,\n",
       " 1235,\n",
       " 1236,\n",
       " 1243,\n",
       " 1260,\n",
       " 1266,\n",
       " 1286,\n",
       " 1287,\n",
       " 1291,\n",
       " 1296,\n",
       " 1297,\n",
       " 1312,\n",
       " 1313,\n",
       " 1318,\n",
       " 1321,\n",
       " 1352,\n",
       " 1413,\n",
       " 1422,\n",
       " 1434,\n",
       " 1456,\n",
       " 1457,\n",
       " 1464,\n",
       " 1475,\n",
       " 1480,\n",
       " 1494,\n",
       " 1531,\n",
       " 1590,\n",
       " 1604,\n",
       " 1619,\n",
       " 1621,\n",
       " 1625,\n",
       " 1675,\n",
       " 1683,\n",
       " 1700,\n",
       " 1713,\n",
       " 1758,\n",
       " 1772,\n",
       " 1781,\n",
       " 1797,\n",
       " 1802,\n",
       " 1807,\n",
       " 1812,\n",
       " 1839,\n",
       " 1845,\n",
       " 1888,\n",
       " 1896,\n",
       " 1899,\n",
       " 1901,\n",
       " 1906,\n",
       " 1921,\n",
       " 1927,\n",
       " 1945,\n",
       " 1987,\n",
       " 1994,\n",
       " 2007,\n",
       " 2019,\n",
       " 2030,\n",
       " 2052,\n",
       " 2054,\n",
       " 2066,\n",
       " 2097,\n",
       " 2103,\n",
       " 2113,\n",
       " 2122,\n",
       " 2123,\n",
       " 2128,\n",
       " 2130,\n",
       " 2138,\n",
       " 2144,\n",
       " 2146,\n",
       " 2147,\n",
       " 2158,\n",
       " 2164,\n",
       " 2190,\n",
       " 2192,\n",
       " 2198,\n",
       " 2205,\n",
       " 2208,\n",
       " 2225,\n",
       " 2232,\n",
       " 2258,\n",
       " 2277,\n",
       " 2293,\n",
       " 2341,\n",
       " 2356,\n",
       " 2361,\n",
       " 2367,\n",
       " 2378,\n",
       " 2405,\n",
       " 2445,\n",
       " 2455,\n",
       " 2463,\n",
       " 2466,\n",
       " 2473,\n",
       " 2481,\n",
       " 2488,\n",
       " 2494,\n",
       " 2510,\n",
       " 2511,\n",
       " 2519,\n",
       " 2527,\n",
       " 2534,\n",
       " 2537,\n",
       " 2543,\n",
       " 2576,\n",
       " 2579,\n",
       " 2581,\n",
       " 2621,\n",
       " 2632,\n",
       " 2640,\n",
       " 2642,\n",
       " 2648,\n",
       " 2649,\n",
       " 2652,\n",
       " 2663,\n",
       " 2665,\n",
       " 2668,\n",
       " 2670,\n",
       " 2719,\n",
       " 2723,\n",
       " 2739,\n",
       " 2752,\n",
       " 2776,\n",
       " 2778,\n",
       " 2781,\n",
       " 2782,\n",
       " 2786,\n",
       " 2825,\n",
       " 2833,\n",
       " 2849,\n",
       " 2850,\n",
       " 2857,\n",
       " 2892,\n",
       " 2897,\n",
       " 2899,\n",
       " 2911,\n",
       " 2926,\n",
       " 2927,\n",
       " 2932,\n",
       " 2943,\n",
       " 2948,\n",
       " 2949,\n",
       " 2964,\n",
       " 2974,\n",
       " 2977,\n",
       " 2985,\n",
       " 3010,\n",
       " 3015,\n",
       " 3021,\n",
       " 3034,\n",
       " 3050,\n",
       " 3052,\n",
       " 3058,\n",
       " 3063,\n",
       " 3064,\n",
       " 3105,\n",
       " 3135,\n",
       " 3149,\n",
       " 3166,\n",
       " 3167,\n",
       " 3176,\n",
       " 3193,\n",
       " 3194,\n",
       " 3197,\n",
       " 3222,\n",
       " 3223,\n",
       " 3236,\n",
       " 3245,\n",
       " 3275,\n",
       " 3282,\n",
       " 3283,\n",
       " 3296,\n",
       " 3300,\n",
       " 3301,\n",
       " 3306,\n",
       " 3316,\n",
       " 3340,\n",
       " 3347,\n",
       " 3350,\n",
       " 3356,\n",
       " 3385,\n",
       " 3410,\n",
       " 3422,\n",
       " 3442,\n",
       " 3459,\n",
       " 3471,\n",
       " 3476,\n",
       " 3483,\n",
       " 3511,\n",
       " 3524,\n",
       " 3533,\n",
       " 3535,\n",
       " 3556,\n",
       " 3573,\n",
       " 3581,\n",
       " 3582,\n",
       " 3586,\n",
       " 3628,\n",
       " 3636,\n",
       " 3644,\n",
       " 3656,\n",
       " 3660,\n",
       " 3670,\n",
       " 3674,\n",
       " 3677,\n",
       " 3681,\n",
       " 3689,\n",
       " 3695,\n",
       " 3717,\n",
       " 3720,\n",
       " 3723,\n",
       " 3739,\n",
       " 3741,\n",
       " 3747,\n",
       " 3767,\n",
       " 3795,\n",
       " 3801,\n",
       " 3807,\n",
       " 3812,\n",
       " 3823,\n",
       " 3832,\n",
       " 3850,\n",
       " 3869,\n",
       " 3870,\n",
       " 3886,\n",
       " 3890,\n",
       " 3893,\n",
       " 3910,\n",
       " 3920,\n",
       " 3922,\n",
       " 3928,\n",
       " 3946,\n",
       " 3960,\n",
       " 4013,\n",
       " 4035,\n",
       " 4045,\n",
       " 4047,\n",
       " 4051,\n",
       " 4065,\n",
       " 4069,\n",
       " 4072,\n",
       " 4078,\n",
       " 4084,\n",
       " 4088,\n",
       " 4097,\n",
       " 4125,\n",
       " 4137,\n",
       " 4164,\n",
       " 4169,\n",
       " 4177,\n",
       " 4178,\n",
       " 4187,\n",
       " 4200,\n",
       " 4202,\n",
       " 4204,\n",
       " 4226,\n",
       " 4246,\n",
       " 4262,\n",
       " 4268,\n",
       " 4276,\n",
       " 4282,\n",
       " 4284,\n",
       " 4297,\n",
       " 4327,\n",
       " 4346,\n",
       " 4350,\n",
       " 4364,\n",
       " 4387,\n",
       " 4403,\n",
       " 4407,\n",
       " 4409,\n",
       " 4428,\n",
       " 4446,\n",
       " 4447,\n",
       " 4451,\n",
       " 4458,\n",
       " 4463,\n",
       " 4465,\n",
       " 4498,\n",
       " 4508,\n",
       " 4515,\n",
       " 4548,\n",
       " 4562,\n",
       " 4594,\n",
       " 4598,\n",
       " 4626,\n",
       " 4640,\n",
       " 4645,\n",
       " 4660,\n",
       " 4702,\n",
       " 4715,\n",
       " 4726,\n",
       " 4728,\n",
       " 4768,\n",
       " 4769,\n",
       " 4805,\n",
       " 4808,\n",
       " 4817,\n",
       " 4840,\n",
       " 4869,\n",
       " 4894,\n",
       " 4898,\n",
       " 4905,\n",
       " 4919,\n",
       " 4924,\n",
       " 4926,\n",
       " 4932,\n",
       " 4935,\n",
       " 4997,\n",
       " 4999,\n",
       " 5026,\n",
       " 5029,\n",
       " 5034,\n",
       " 5036,\n",
       " 5045,\n",
       " 5047,\n",
       " 5059,\n",
       " 5077,\n",
       " 5090,\n",
       " 5095,\n",
       " 5103,\n",
       " 5109,\n",
       " 5112,\n",
       " 5116,\n",
       " 5118,\n",
       " 5137,\n",
       " 5139,\n",
       " 5153,\n",
       " 5174,\n",
       " 5177,\n",
       " 5202,\n",
       " 5206,\n",
       " 5210,\n",
       " 5216,\n",
       " 5238,\n",
       " 5251,\n",
       " 5271,\n",
       " 5274,\n",
       " 5349,\n",
       " 5381,\n",
       " 5393,\n",
       " 5406,\n",
       " 5407,\n",
       " 5417,\n",
       " 5425,\n",
       " 5442,\n",
       " 5453,\n",
       " 5474,\n",
       " 5484,\n",
       " 5492,\n",
       " 5493,\n",
       " 5523,\n",
       " 5549,\n",
       " 5551,\n",
       " 5565,\n",
       " 5575,\n",
       " 5583,\n",
       " 5587,\n",
       " 5604,\n",
       " 5606,\n",
       " 5617,\n",
       " 5639,\n",
       " 5653,\n",
       " 5655,\n",
       " 5664,\n",
       " 5672,\n",
       " 5718,\n",
       " 5720,\n",
       " 5728,\n",
       " 5745,\n",
       " 5752,\n",
       " 5756,\n",
       " 5767,\n",
       " 5768,\n",
       " 5783,\n",
       " 5813,\n",
       " 5838,\n",
       " 5843,\n",
       " 5856,\n",
       " 5894,\n",
       " 5903,\n",
       " 5921,\n",
       " 5925,\n",
       " 5926,\n",
       " 5963,\n",
       " 5988,\n",
       " 5996,\n",
       " 5999,\n",
       " 6007,\n",
       " 6043,\n",
       " 6044,\n",
       " 6054,\n",
       " 6068,\n",
       " 6075,\n",
       " 6076,\n",
       " 6091,\n",
       " 6093,\n",
       " 6103,\n",
       " 6116,\n",
       " 6118,\n",
       " 6131,\n",
       " 6134,\n",
       " 6145,\n",
       " 6168,\n",
       " 6175,\n",
       " 6186,\n",
       " 6190,\n",
       " 6229,\n",
       " 6232,\n",
       " 6235,\n",
       " 6240,\n",
       " 6249,\n",
       " 6257,\n",
       " 6273,\n",
       " 6276,\n",
       " 6327,\n",
       " 6329,\n",
       " 6331,\n",
       " 6334,\n",
       " 6337,\n",
       " 6359,\n",
       " 6367,\n",
       " 6372,\n",
       " 6380,\n",
       " 6381,\n",
       " 6407,\n",
       " 6421,\n",
       " 6453,\n",
       " 6458,\n",
       " 6462,\n",
       " 6464,\n",
       " 6467,\n",
       " 6522,\n",
       " 6561,\n",
       " 6589,\n",
       " 6594,\n",
       " 6603,\n",
       " 6606,\n",
       " 6613,\n",
       " 6614,\n",
       " 6634,\n",
       " 6654,\n",
       " 6656,\n",
       " 6691,\n",
       " 6700,\n",
       " 6732,\n",
       " 6736,\n",
       " 6747,\n",
       " 6750,\n",
       " 6771,\n",
       " 6785,\n",
       " 6792,\n",
       " 6799,\n",
       " 6828,\n",
       " 6829,\n",
       " 6832,\n",
       " 6840,\n",
       " 6843,\n",
       " 6854,\n",
       " 6872,\n",
       " 6895,\n",
       " 6901,\n",
       " 6907,\n",
       " 6918,\n",
       " 6931,\n",
       " 6935,\n",
       " 6937,\n",
       " 6970,\n",
       " 6971,\n",
       " 6980,\n",
       " 6989,\n",
       " 6991,\n",
       " 6993,\n",
       " 7006,\n",
       " 7030,\n",
       " 7061,\n",
       " 7064,\n",
       " 7068,\n",
       " 7069,\n",
       " 7098,\n",
       " 7100,\n",
       " 7102,\n",
       " 7108,\n",
       " 7113,\n",
       " 7114,\n",
       " 7129,\n",
       " 7172,\n",
       " 7175,\n",
       " 7190,\n",
       " 7234,\n",
       " 7235,\n",
       " 7242,\n",
       " 7273,\n",
       " 7277,\n",
       " 7303,\n",
       " 7308,\n",
       " 7309,\n",
       " 7313,\n",
       " 7317,\n",
       " 7330,\n",
       " 7343,\n",
       " 7350,\n",
       " 7376,\n",
       " 7377,\n",
       " 7453,\n",
       " 7460,\n",
       " 7466,\n",
       " 7473,\n",
       " 7480,\n",
       " 7483,\n",
       " 7501,\n",
       " 7503,\n",
       " 7523,\n",
       " 7528,\n",
       " 7538,\n",
       " 7627,\n",
       " 7644,\n",
       " 7645,\n",
       " 7663,\n",
       " 7677,\n",
       " 7682,\n",
       " 7691,\n",
       " 7755,\n",
       " 7768,\n",
       " 7774,\n",
       " 7788,\n",
       " 7795,\n",
       " 7802,\n",
       " 7804,\n",
       " 7821,\n",
       " 7834,\n",
       " 7835,\n",
       " 7847,\n",
       " 7863,\n",
       " 7865,\n",
       " 7885,\n",
       " 7890,\n",
       " 7891,\n",
       " 7901,\n",
       " 7938,\n",
       " 7950,\n",
       " 7951,\n",
       " 7955,\n",
       " 7962,\n",
       " 7971,\n",
       " 7980,\n",
       " 7984,\n",
       " 7989,\n",
       " 7995,\n",
       " 8042,\n",
       " 8058,\n",
       " 8073,\n",
       " 8091,\n",
       " 8094,\n",
       " 8116,\n",
       " 8132,\n",
       " 8141,\n",
       " 8172,\n",
       " 8174,\n",
       " 8235,\n",
       " 8295,\n",
       " 8301,\n",
       " 8325,\n",
       " 8335,\n",
       " 8337,\n",
       " 8373,\n",
       " 8379,\n",
       " 8405,\n",
       " 8409,\n",
       " 8413,\n",
       " 8414,\n",
       " 8415,\n",
       " 8431,\n",
       " 8434,\n",
       " 8437,\n",
       " 8467,\n",
       " 8479,\n",
       " 8486,\n",
       " 8516,\n",
       " 8544,\n",
       " 8547,\n",
       " 8560,\n",
       " 8578,\n",
       " 8579,\n",
       " 8590,\n",
       " 8594,\n",
       " 8596,\n",
       " 8597,\n",
       " 8604,\n",
       " 8606,\n",
       " 8612,\n",
       " 8614,\n",
       " 8620,\n",
       " 8642,\n",
       " 8643,\n",
       " 8653,\n",
       " 8670,\n",
       " 8679,\n",
       " 8685,\n",
       " 8687,\n",
       " 8691,\n",
       " 8692,\n",
       " 8696,\n",
       " 8708,\n",
       " 8714,\n",
       " 8722,\n",
       " 8730,\n",
       " 8732,\n",
       " 8733,\n",
       " 8738,\n",
       " 8739,\n",
       " 8769,\n",
       " 8778,\n",
       " 8779,\n",
       " 8808,\n",
       " 8838,\n",
       " 8871,\n",
       " 8873,\n",
       " 8881,\n",
       " 8886,\n",
       " 8891,\n",
       " 8897,\n",
       " 8932,\n",
       " 8944,\n",
       " 8958,\n",
       " 8978,\n",
       " 8994,\n",
       " 9008,\n",
       " 9026,\n",
       " 9068,\n",
       " 9073,\n",
       " 9074,\n",
       " 9093,\n",
       " 9099,\n",
       " 9122,\n",
       " 9134,\n",
       " 9144,\n",
       " 9173,\n",
       " 9188,\n",
       " 9195,\n",
       " 9199,\n",
       " 9206,\n",
       " 9215,\n",
       " 9222,\n",
       " 9232,\n",
       " 9260,\n",
       " 9261,\n",
       " 9262,\n",
       " 9275,\n",
       " 9290,\n",
       " 9309,\n",
       " 9310,\n",
       " 9314,\n",
       " 9316,\n",
       " 9318,\n",
       " 9321,\n",
       " 9322,\n",
       " 9361,\n",
       " 9399,\n",
       " 9400,\n",
       " 9407,\n",
       " 9416,\n",
       " 9429,\n",
       " 9461,\n",
       " 9477,\n",
       " 9483,\n",
       " 9500,\n",
       " 9507,\n",
       " 9514,\n",
       " 9522,\n",
       " 9524,\n",
       " 9531,\n",
       " 9546,\n",
       " 9557,\n",
       " 9582,\n",
       " 9583,\n",
       " 9602,\n",
       " 9609,\n",
       " 9621,\n",
       " 9623,\n",
       " 9624,\n",
       " 9627,\n",
       " 9633,\n",
       " 9671,\n",
       " 9679,\n",
       " 9687,\n",
       " 9711,\n",
       " 9749,\n",
       " 9767,\n",
       " 9775,\n",
       " 9782,\n",
       " 9798,\n",
       " 9827,\n",
       " 9844,\n",
       " 9857,\n",
       " 9870,\n",
       " 9880,\n",
       " 9956,\n",
       " 9963,\n",
       " 9974,\n",
       " 9992,\n",
       " 9993,\n",
       " 10032,\n",
       " 10033,\n",
       " 10042,\n",
       " 10068,\n",
       " 10069,\n",
       " 10082,\n",
       " 10098,\n",
       " 10103,\n",
       " 10116,\n",
       " 10118,\n",
       " 10155,\n",
       " 10181,\n",
       " 10228,\n",
       " 10234,\n",
       " 10239,\n",
       " 10246,\n",
       " 10248,\n",
       " 10276,\n",
       " 10281,\n",
       " 10294,\n",
       " 10304,\n",
       " 10307,\n",
       " 10350,\n",
       " 10356,\n",
       " 10366,\n",
       " 10367,\n",
       " 10372,\n",
       " 10399,\n",
       " 10411,\n",
       " 10423,\n",
       " 10428,\n",
       " 10429,\n",
       " 10434,\n",
       " 10437,\n",
       " 10448,\n",
       " 10474,\n",
       " 10476,\n",
       " 10528,\n",
       " 10535,\n",
       " 10543,\n",
       " 10570,\n",
       " 10595,\n",
       " 10604,\n",
       " 10614,\n",
       " 10617,\n",
       " 10625,\n",
       " 10626,\n",
       " 10647,\n",
       " 10658,\n",
       " 10668,\n",
       " 10683,\n",
       " 10710,\n",
       " 10711,\n",
       " 10716,\n",
       " 10734,\n",
       " 10761,\n",
       " 10805,\n",
       " 10826,\n",
       " 10832,\n",
       " 10836,\n",
       " 10844,\n",
       " 10873,\n",
       " 10876,\n",
       " 10883,\n",
       " 10903,\n",
       " 10910,\n",
       " 10926,\n",
       " 10936,\n",
       " 10943,\n",
       " 10996,\n",
       " 11024,\n",
       " 11028,\n",
       " 11057,\n",
       " 11060,\n",
       " 11064,\n",
       " 11067,\n",
       " 11071,\n",
       " 11091,\n",
       " 11103,\n",
       " 11107,\n",
       " 11125,\n",
       " 11137,\n",
       " 11139,\n",
       " 11144,\n",
       " 11158,\n",
       " 11167,\n",
       " 11207,\n",
       " 11242,\n",
       " 11270,\n",
       " 11278,\n",
       " 11291,\n",
       " 11302,\n",
       " 11305,\n",
       " 11322,\n",
       " 11324,\n",
       " 11329,\n",
       " 11344,\n",
       " 11353,\n",
       " 11359,\n",
       " 11363,\n",
       " 11375,\n",
       " 11411,\n",
       " 11428,\n",
       " 11444,\n",
       " 11452,\n",
       " 11476,\n",
       " 11484,\n",
       " 11498,\n",
       " 11509,\n",
       " 11517,\n",
       " 11527,\n",
       " 11562,\n",
       " 11574,\n",
       " 11619,\n",
       " 11640,\n",
       " 11671,\n",
       " 11674,\n",
       " 11695,\n",
       " 11710,\n",
       " 11717,\n",
       " 11731,\n",
       " 11735,\n",
       " 11756,\n",
       " 11767,\n",
       " 11788,\n",
       " 11814,\n",
       " 11826,\n",
       " 11842,\n",
       " 11846,\n",
       " 11862,\n",
       " 11867,\n",
       " 11884,\n",
       " 11894,\n",
       " 11906,\n",
       " 11914,\n",
       " 11925,\n",
       " 11927,\n",
       " 11938,\n",
       " 11952,\n",
       " 11958,\n",
       " 11959,\n",
       " 11980,\n",
       " 12012,\n",
       " 12025,\n",
       " 12045,\n",
       " 12048,\n",
       " 12066,\n",
       " 12069,\n",
       " 12071,\n",
       " 12083,\n",
       " 12121,\n",
       " 12132,\n",
       " 12135,\n",
       " 12145,\n",
       " 12163,\n",
       " 12184,\n",
       " 12185,\n",
       " 12191,\n",
       " 12207,\n",
       " 12241,\n",
       " 12249,\n",
       " 12257,\n",
       " 12259,\n",
       " 12260,\n",
       " 12267,\n",
       " 12281,\n",
       " 12295,\n",
       " 12297,\n",
       " 12301,\n",
       " 12327,\n",
       " 12328,\n",
       " 12338,\n",
       " 12342,\n",
       " 12347,\n",
       " 12355,\n",
       " 12400,\n",
       " 12457,\n",
       " 12462,\n",
       " 12473,\n",
       " 12486,\n",
       " 12489,\n",
       " 12498,\n",
       " 12505,\n",
       " 12506,\n",
       " 12516,\n",
       " 12518,\n",
       " 12521,\n",
       " 12539,\n",
       " 12541,\n",
       " 12542,\n",
       " 12576,\n",
       " 12581,\n",
       " 12589,\n",
       " ...]"
      ]
     },
     "execution_count": 273,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "unique_reaction_subset = sorted(chosen_reactions)\n",
    "unique_reaction_subset\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 258,
   "metadata": {},
   "outputs": [],
   "source": [
    "unique_reactions = data.iloc[unique_reaction_subset]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 259,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>yieldclass</th>\n",
       "      <th>Component_1</th>\n",
       "      <th>Component_2</th>\n",
       "      <th>Product</th>\n",
       "      <th>property_P</th>\n",
       "      <th>property_C1</th>\n",
       "      <th>property_C2</th>\n",
       "      <th>MolLogP_(C1_rdkit)</th>\n",
       "      <th>MolMR_(C1_rdkit)</th>\n",
       "      <th>MolWt_(C1_rdkit)</th>\n",
       "      <th>...</th>\n",
       "      <th>MQN33_(P_rdkit)</th>\n",
       "      <th>MQN34_(P_rdkit)</th>\n",
       "      <th>MQN35_(P_rdkit)</th>\n",
       "      <th>MQN36_(P_rdkit)</th>\n",
       "      <th>MQN37_(P_rdkit)</th>\n",
       "      <th>MQN38_(P_rdkit)</th>\n",
       "      <th>MQN39_(P_rdkit)</th>\n",
       "      <th>MQN40_(P_rdkit)</th>\n",
       "      <th>MQN41_(P_rdkit)</th>\n",
       "      <th>MQN42_(P_rdkit)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>CN(C)C1=CC=C(Br)C=C1C(F)(F)F</td>\n",
       "      <td>OB(O)C1=CC(OCC2=C(F)C=CC=C2)=CC=C1</td>\n",
       "      <td>CN(C)C1=C(C=C(C=C1)C1=CC(OCC2=C(F)C=CC=C2)=CC=...</td>\n",
       "      <td>21.316</td>\n",
       "      <td>9.286395</td>\n",
       "      <td>4.779728</td>\n",
       "      <td>3.53390</td>\n",
       "      <td>53.4710</td>\n",
       "      <td>268.076</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>0</td>\n",
       "      <td>CC(=O)NC1=C(Br)C=C(F)C=C1</td>\n",
       "      <td>OB(C1=CC(Cl)=C(Cl)C=C1)C1=CC=C(Cl)C(Cl)=C1</td>\n",
       "      <td>CC(=O)NC1=C(C=C(F)C=C1)C1=CC(Cl)=C(Cl)C=C1</td>\n",
       "      <td>3.148</td>\n",
       "      <td>10.706549</td>\n",
       "      <td>4.705935</td>\n",
       "      <td>2.54660</td>\n",
       "      <td>48.4037</td>\n",
       "      <td>232.052</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>0</td>\n",
       "      <td>CC\\C(COCC1=CC=C(OC)C=C1)=C/I</td>\n",
       "      <td>CC[C@@H](\\C=C(/CC)CO)[C@@H](OCOC)[C@H](C)\\C=C(...</td>\n",
       "      <td>CC[C@@H](\\C=C(/CC)C=O)[C@@H](OCOC)[C@H](C)\\C=C...</td>\n",
       "      <td>20.997</td>\n",
       "      <td>10.298637</td>\n",
       "      <td>4.512120</td>\n",
       "      <td>3.94070</td>\n",
       "      <td>75.0440</td>\n",
       "      <td>332.181</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>0</td>\n",
       "      <td>NC1=CC=C(Cl)N=C1I</td>\n",
       "      <td>OB(O)C1=CC2=C(F)C=CC=C2N1</td>\n",
       "      <td>NC1=CC=C(Cl)N=C1C1=CC2=C(F)C=CC=C2N1</td>\n",
       "      <td>42.677</td>\n",
       "      <td>10.592522</td>\n",
       "      <td>4.804849</td>\n",
       "      <td>1.92180</td>\n",
       "      <td>46.3764</td>\n",
       "      <td>254.458</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37</th>\n",
       "      <td>0</td>\n",
       "      <td>ClC1=CC=C(C=C1)C(=O)C1=CC=CC=C1</td>\n",
       "      <td>CC(=O)OC[B-](F)(F)F</td>\n",
       "      <td>OCC1=CC=C(C=C1)C(=O)C1=CC=CC=C1</td>\n",
       "      <td>46.493</td>\n",
       "      <td>10.859668</td>\n",
       "      <td>4.742911</td>\n",
       "      <td>3.57100</td>\n",
       "      <td>61.3265</td>\n",
       "      <td>216.667</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>54260</th>\n",
       "      <td>0</td>\n",
       "      <td>CC1=CC2=CC(=CC(Br)=C2O1)C#N</td>\n",
       "      <td>CC(C)(C)OC(=O)C1=C(O)C=C(C=C1)B1OC(C)(C)C(C)(C)O1</td>\n",
       "      <td>CC1=CC2=CC(=CC(=C2O1)C1=CC(O)=C(C=C1)C(=O)OC(C...</td>\n",
       "      <td>10.097</td>\n",
       "      <td>10.178396</td>\n",
       "      <td>4.410485</td>\n",
       "      <td>3.37540</td>\n",
       "      <td>53.3660</td>\n",
       "      <td>236.068</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>54263</th>\n",
       "      <td>1</td>\n",
       "      <td>CC(=O)N1CCC2=C(C1)C(I)=NN2CC1CC1</td>\n",
       "      <td>CC1(C)OB(OC1(C)C)C1=CC=CC2=C1C=CN2</td>\n",
       "      <td>CC(=O)N1CCC2=C(C1)C(=NN2CC1CC1)C1=CC=CC2=C1C=CN2</td>\n",
       "      <td>3.078</td>\n",
       "      <td>10.661869</td>\n",
       "      <td>4.634257</td>\n",
       "      <td>1.80230</td>\n",
       "      <td>72.4630</td>\n",
       "      <td>345.184</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>54266</th>\n",
       "      <td>1</td>\n",
       "      <td>CNC(=O)C1=C(Br)C=CC=N1</td>\n",
       "      <td>CC1(C)OB(OC1(C)C)C1=CC=C(CCC(=O)NCCNS(=O)(=O)C...</td>\n",
       "      <td>CNC(=O)C1=C(C=CC=N1)C1=CC=C(CCC(=O)NCCNS(=O)(=...</td>\n",
       "      <td>23.572</td>\n",
       "      <td>9.797723</td>\n",
       "      <td>4.803503</td>\n",
       "      <td>1.20370</td>\n",
       "      <td>45.3272</td>\n",
       "      <td>215.050</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>54274</th>\n",
       "      <td>0</td>\n",
       "      <td>CC(=O)C1=NC(Br)=CC=C1</td>\n",
       "      <td>OB(O)C1=C(O)C=C(Cl)C=C1</td>\n",
       "      <td>CC(=O)C1=NC(=CC=C1)C1=C(O)C=CC(Cl)=C1</td>\n",
       "      <td>3.027</td>\n",
       "      <td>10.286742</td>\n",
       "      <td>4.847903</td>\n",
       "      <td>2.04670</td>\n",
       "      <td>41.9415</td>\n",
       "      <td>200.035</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>54275</th>\n",
       "      <td>0</td>\n",
       "      <td>ClC1=CC=C(C=C1)C#N</td>\n",
       "      <td>F[B-](F)(F)C1=CC2=CC=CC=C2S1</td>\n",
       "      <td>N#CC1=CC=C(C=C1)C1=CC2=C(S1)C=CC=C2</td>\n",
       "      <td>41.480</td>\n",
       "      <td>10.950262</td>\n",
       "      <td>4.781281</td>\n",
       "      <td>2.21168</td>\n",
       "      <td>36.1670</td>\n",
       "      <td>137.569</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>4345 rows × 463 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       yieldclass                       Component_1  \\\n",
       "4               0      CN(C)C1=CC=C(Br)C=C1C(F)(F)F   \n",
       "9               0         CC(=O)NC1=C(Br)C=C(F)C=C1   \n",
       "23              0      CC\\C(COCC1=CC=C(OC)C=C1)=C/I   \n",
       "32              0                 NC1=CC=C(Cl)N=C1I   \n",
       "37              0   ClC1=CC=C(C=C1)C(=O)C1=CC=CC=C1   \n",
       "...           ...                               ...   \n",
       "54260           0       CC1=CC2=CC(=CC(Br)=C2O1)C#N   \n",
       "54263           1  CC(=O)N1CCC2=C(C1)C(I)=NN2CC1CC1   \n",
       "54266           1            CNC(=O)C1=C(Br)C=CC=N1   \n",
       "54274           0             CC(=O)C1=NC(Br)=CC=C1   \n",
       "54275           0                ClC1=CC=C(C=C1)C#N   \n",
       "\n",
       "                                             Component_2  \\\n",
       "4                     OB(O)C1=CC(OCC2=C(F)C=CC=C2)=CC=C1   \n",
       "9             OB(C1=CC(Cl)=C(Cl)C=C1)C1=CC=C(Cl)C(Cl)=C1   \n",
       "23     CC[C@@H](\\C=C(/CC)CO)[C@@H](OCOC)[C@H](C)\\C=C(...   \n",
       "32                             OB(O)C1=CC2=C(F)C=CC=C2N1   \n",
       "37                                   CC(=O)OC[B-](F)(F)F   \n",
       "...                                                  ...   \n",
       "54260  CC(C)(C)OC(=O)C1=C(O)C=C(C=C1)B1OC(C)(C)C(C)(C)O1   \n",
       "54263                 CC1(C)OB(OC1(C)C)C1=CC=CC2=C1C=CN2   \n",
       "54266  CC1(C)OB(OC1(C)C)C1=CC=C(CCC(=O)NCCNS(=O)(=O)C...   \n",
       "54274                            OB(O)C1=C(O)C=C(Cl)C=C1   \n",
       "54275                       F[B-](F)(F)C1=CC2=CC=CC=C2S1   \n",
       "\n",
       "                                                 Product  property_P  \\\n",
       "4      CN(C)C1=C(C=C(C=C1)C1=CC(OCC2=C(F)C=CC=C2)=CC=...      21.316   \n",
       "9             CC(=O)NC1=C(C=C(F)C=C1)C1=CC(Cl)=C(Cl)C=C1       3.148   \n",
       "23     CC[C@@H](\\C=C(/CC)C=O)[C@@H](OCOC)[C@H](C)\\C=C...      20.997   \n",
       "32                  NC1=CC=C(Cl)N=C1C1=CC2=C(F)C=CC=C2N1      42.677   \n",
       "37                       OCC1=CC=C(C=C1)C(=O)C1=CC=CC=C1      46.493   \n",
       "...                                                  ...         ...   \n",
       "54260  CC1=CC2=CC(=CC(=C2O1)C1=CC(O)=C(C=C1)C(=O)OC(C...      10.097   \n",
       "54263   CC(=O)N1CCC2=C(C1)C(=NN2CC1CC1)C1=CC=CC2=C1C=CN2       3.078   \n",
       "54266  CNC(=O)C1=C(C=CC=N1)C1=CC=C(CCC(=O)NCCNS(=O)(=...      23.572   \n",
       "54274              CC(=O)C1=NC(=CC=C1)C1=C(O)C=CC(Cl)=C1       3.027   \n",
       "54275                N#CC1=CC=C(C=C1)C1=CC2=C(S1)C=CC=C2      41.480   \n",
       "\n",
       "       property_C1  property_C2  MolLogP_(C1_rdkit)  MolMR_(C1_rdkit)  \\\n",
       "4         9.286395     4.779728             3.53390           53.4710   \n",
       "9        10.706549     4.705935             2.54660           48.4037   \n",
       "23       10.298637     4.512120             3.94070           75.0440   \n",
       "32       10.592522     4.804849             1.92180           46.3764   \n",
       "37       10.859668     4.742911             3.57100           61.3265   \n",
       "...            ...          ...                 ...               ...   \n",
       "54260    10.178396     4.410485             3.37540           53.3660   \n",
       "54263    10.661869     4.634257             1.80230           72.4630   \n",
       "54266     9.797723     4.803503             1.20370           45.3272   \n",
       "54274    10.286742     4.847903             2.04670           41.9415   \n",
       "54275    10.950262     4.781281             2.21168           36.1670   \n",
       "\n",
       "       MolWt_(C1_rdkit)  ...  MQN33_(P_rdkit)  MQN34_(P_rdkit)  \\\n",
       "4               268.076  ...                0                0   \n",
       "9               232.052  ...                0                0   \n",
       "23              332.181  ...                0                0   \n",
       "32              254.458  ...                0                0   \n",
       "37              216.667  ...                0                0   \n",
       "...                 ...  ...              ...              ...   \n",
       "54260           236.068  ...                0                0   \n",
       "54263           345.184  ...                1                0   \n",
       "54266           215.050  ...                0                0   \n",
       "54274           200.035  ...                0                0   \n",
       "54275           137.569  ...                0                0   \n",
       "\n",
       "       MQN35_(P_rdkit)  MQN36_(P_rdkit)  MQN37_(P_rdkit)  MQN38_(P_rdkit)  \\\n",
       "4                    0                3                0                0   \n",
       "9                    0                2                0                0   \n",
       "23                   0                1                0                0   \n",
       "32                   1                2                0                0   \n",
       "37                   0                2                0                0   \n",
       "...                ...              ...              ...              ...   \n",
       "54260                1                2                0                0   \n",
       "54263                2                2                0                0   \n",
       "54266                0                4                0                0   \n",
       "54274                0                2                0                0   \n",
       "54275                1                2                0                0   \n",
       "\n",
       "       MQN39_(P_rdkit)  MQN40_(P_rdkit)  MQN41_(P_rdkit)  MQN42_(P_rdkit)  \n",
       "4                    0                0                0                0  \n",
       "9                    0                0                0                0  \n",
       "23                   0                0                0                0  \n",
       "32                   0                0                2                1  \n",
       "37                   0                0                0                0  \n",
       "...                ...              ...              ...              ...  \n",
       "54260                0                0                2                1  \n",
       "54263                0                0                4                2  \n",
       "54266                0                0                2                1  \n",
       "54274                0                0                0                0  \n",
       "54275                0                0                2                1  \n",
       "\n",
       "[4345 rows x 463 columns]"
      ]
     },
     "execution_count": 259,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "unique_reactions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 228,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 75938/75938 [00:16<00:00, 4556.35it/s]\n"
     ]
    }
   ],
   "source": [
    "u_molecules1 = set(unique_reactions['Component_1'].unique())\n",
    "u_molecules2 = set(unique_reactions['Component_2'].unique())\n",
    "u_products = set(unique_reactions['Product'].unique())\n",
    "u_all_mols = molecules1 | molecules2 | products\n",
    "u_cleaned_mols = list(set(canonicalize_smiles(smiles) for smiles in tqdm(u_all_mols)))\n",
    "np.random.shuffle(u_cleaned_mols)\n",
    "u_mol_id = {mol: i for i,mol in enumerate(u_cleaned_mols)}\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 229,
   "metadata": {},
   "outputs": [],
   "source": [
    "u_mol_counts = Counter()\n",
    "u_mol_counts.update(unique_reactions['Component_1'])\n",
    "u_mol_counts.update(unique_reactions['Component_2'])\n",
    "u_mol_counts.update(unique_reactions['Product'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 230,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('ClC1=CC=CC(I)=C1', 2),\n",
       " ('BrC1=CC=C(S1)C1=CC=C(Br)S1', 2),\n",
       " ('CC(=O)NC1=NC2=C(S1)C=C(Br)C=C2', 2),\n",
       " ('[O-][N+](=O)C1=C(Br)C=CC=C1', 2),\n",
       " ('BrC1=CC=C(C=C1)C1=CN(N=N1)C(=O)N1CCCCC1CC1=CC=CC=C1', 2),\n",
       " ('COC1=CC=C(CN(C2CCCC2)S(=O)(=O)C2=CC(Br)=CC=C2OC)C(OC)=C1', 2),\n",
       " ('CC1=CC=C(C=C1)S(=O)(=O)N1C=CC2=C(Cl)N=CN=C12', 2),\n",
       " ('BrC1=CC=CS1', 2),\n",
       " ('CS(=O)(=O)C1=C(NC2=CC(Cl)=C3N=CC=NC3=C2)C=NC=C1', 2),\n",
       " ('CC1=CC=C(O)C(Br)=C1', 2)]"
      ]
     },
     "execution_count": 230,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "u_mol_counts.most_common(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "5254"
      ]
     },
     "execution_count": 138,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(unique_reactions)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 260,
   "metadata": {},
   "outputs": [],
   "source": [
    "u_y = unique_reactions['yieldclass'].values\n",
    "u_properties = unique_reactions.drop(columns=['yieldclass', 'Component_1', 'Component_2', 'Product', 'property_C1', 'property_C2', 'property_P'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 261,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "9c84ad8aaeb347119ee08488423e8b7f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "MetricVisualizer(layout=Layout(align_self='stretch', height='500px'))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<catboost.core.CatBoostClassifier at 0x1fc43e378e0>"
      ]
     },
     "execution_count": 261,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from catboost import CatBoostClassifier, Pool, metrics, cv\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "rng = np.random.default_rng(1729)\n",
    "train_indices, test_indices = split_train_test_reactions(reaction_components, rng=rng)\n",
    "#X_train, X_validation, y_train, y_validation = train_test_split(u_properties.values, u_y, train_size=0.8, random_state=42)\n",
    "\n",
    "X_train = symbolic_reactions\n",
    "\n",
    "model = CatBoostClassifier(\n",
    "    #custom_loss=[metrics.Accuracy()],\n",
    "    custom_metric=[metrics.Accuracy(), metrics.BalancedAccuracy(), metrics.AUC(), metrics.Precision(), metrics.Recall()],\n",
    "    random_seed=42,\n",
    "    logging_level='Silent'\n",
    ")\n",
    "cv_params = model.get_params()\n",
    "cv_params.update({\n",
    "    'loss_function': metrics.Logloss(),\n",
    "    'iterations': 1000,\n",
    "    'learning_rate': 1e-2\n",
    "})\n",
    "# cv_data = cv(\n",
    "#     Pool(u_properties.values, u_y),\n",
    "#     cv_params,\n",
    "#     plot=True,\n",
    "#     stratified=True,\n",
    "#     fold_count=5,\n",
    "# )\n",
    "model.fit(X_train, y_train, eval_set=(X_validation, y_validation), plot=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 262,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Replace each molecule with its unique integer index\n",
    "u_symbolic_reactions = unique_reactions[['Component_1', 'Component_2', 'Product']].applymap(lambda x: u_mol_id[canonicalize_smiles(x)]).values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 263,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[72011,  6185, 67607],\n",
       "       [59139, 21493, 16814],\n",
       "       [23424, 34902, 29682],\n",
       "       ...,\n",
       "       [ 1965, 69632, 71081],\n",
       "       [71107, 69143,   625],\n",
       "       [42369,  1784, 23963]], dtype=int64)"
      ]
     },
     "execution_count": 263,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "u_symbolic_reactions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 264,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([    3,     7,    11, ..., 75862, 75865, 75870], dtype=int64),\n",
       " array([1, 1, 1, ..., 1, 1, 1], dtype=int64))"
      ]
     },
     "execution_count": 264,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.unique(u_symbolic_reactions, return_counts=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 265,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "abb3fe908f024761aaf8432be483be17",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "MetricVisualizer(layout=Layout(align_self='stretch', height='500px'))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<catboost.core.CatBoostClassifier at 0x1fbf65de460>"
      ]
     },
     "execution_count": 265,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from catboost import CatBoostClassifier, Pool, metrics, cv\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "X_train, X_validation, y_train, y_validation = train_test_split(u_symbolic_reactions, u_y, train_size=0.75, random_state=42)\n",
    "\n",
    "catergorical_feature_indices = [0,1,2]\n",
    "model = CatBoostClassifier(\n",
    "    #custom_loss=[metrics.Accuracy()],\n",
    "    custom_metric=[metrics.Accuracy(), metrics.BalancedAccuracy(), metrics.AUC(), metrics.Precision(), metrics.Recall()],\n",
    "    random_seed=42,\n",
    "    logging_level='Silent',\n",
    "    auto_class_weights='Balanced'\n",
    ")\n",
    "cv_params = model.get_params()\n",
    "cv_params.update({\n",
    "    'loss_function': metrics.Logloss()\n",
    "})\n",
    "# cv_data = cv(\n",
    "#     Pool(u_symbolic_reactions, u_y, cat_features=catergorical_feature_indices),\n",
    "#     cv_params,\n",
    "#     plot=True,\n",
    "#     stratified=True,\n",
    "#     fold_count=3\n",
    "# )\n",
    "model.fit(X_train, y_train, cat_features=catergorical_feature_indices, eval_set=(X_validation, y_validation), plot=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 196,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1296x1296 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "importances = model.get_feature_importance()\n",
    "sort_order = np.argsort(importances)\n",
    "\n",
    "top_k = 30\n",
    "selected_indices = sort_order[-top_k:]\n",
    "selected_importances = importances[selected_indices]\n",
    "x = np.arange(len(selected_importances))\n",
    "labels = [u_properties.columns[i] for i in selected_indices]\n",
    "plt.figure(figsize=(18,18))\n",
    "plt.barh(x, selected_importances)\n",
    "plt.yticks(x, labels);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "7dd5ff0d4892b3d4800ffb8992589ad9d2408512d64cb0ca86ad6a1a9e3b584d"
  },
  "kernelspec": {
   "display_name": "Python 3.9.13 ('yield-prediction')",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.13"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}