Catch PDBBind analysis errors

BioinfoMachineLearning · Aug 11, 2024 · ef89b5b · ef89b5b
1 parent 2e128f0
commit ef89b5b
Showing 1 changed file with 30 additions and 24 deletions.
diff --git a/notebooks/dataset_interaction_analysis_plotting.ipynb b/notebooks/dataset_interaction_analysis_plotting.ipynb
@@ -221,35 +221,41 @@
     "    for protein_filepath, ligand_filepath in tqdm(\n",
     "        pdbbind_protein_ligand_filepath_pairs, desc=\"Processing PDBBind 2020 set\"\n",
     "    ):\n",
-    "        temp_protein_filepath = create_temp_pdb_with_only_molecule_type_residues(\n",
-    "            protein_filepath, molecule_type=\"protein\"\n",
-    "        )\n",
-    "        ligand_mol = None\n",
     "        try:\n",
-    "            ligand_mol = Chem.MolFromMolFile(ligand_filepath)\n",
-    "        except Exception as e:\n",
-    "            ligand_mol = Chem.MolFromMolFile(ligand_filepath, sanitize=False)\n",
-    "        if ligand_mol is None:\n",
-    "            print(\n",
-    "                f\"Using the `.mol2` file for PDBBind 2020 target {ligand_filepath} failed. We found a `.sdf` file instead and are trying to use that. Be aware that the `.sdf` files from PDBBind 2020 are missing chirality tags, although we will do our best to impute such information automatically using RDKit. Reference: https://www.blopig.com/blog/2021/09/watch-out-when-using-pdbbind.\"\n",
+    "            temp_protein_filepath = create_temp_pdb_with_only_molecule_type_residues(\n",
+    "                protein_filepath, molecule_type=\"protein\"\n",
     "            )\n",
+    "            ligand_mol = None\n",
     "            try:\n",
-    "                ligand_mol = Chem.MolFromMolFile(ligand_filepath.replace(\".mol2\", \".sdf\"))\n",
-    "            except Exception as e:\n",
-    "                ligand_mol = Chem.MolFromMolFile(\n",
-    "                    ligand_filepath.replace(\".mol2\", \".sdf\"), sanitize=False\n",
-    "                )\n",
-    "            try:\n",
-    "                Chem.rdmolops.AssignAtomChiralTagsFromStructure(ligand_mol)\n",
+    "                ligand_mol = Chem.MolFromMolFile(ligand_filepath)\n",
     "            except Exception as e:\n",
+    "                ligand_mol = Chem.MolFromMolFile(ligand_filepath, sanitize=False)\n",
+    "            if ligand_mol is None:\n",
     "                print(\n",
-    "                    f\"Could not assign chirality tags to the atoms in the PDBBind ligand molecule from {ligand_filepath}.\"\n",
+    "                    f\"Using the `.mol2` file for PDBBind 2020 target {ligand_filepath} failed. We found a `.sdf` file instead and are trying to use that. Be aware that the `.sdf` files from PDBBind 2020 are missing chirality tags, although we will do our best to impute such information automatically using RDKit. Reference: https://www.blopig.com/blog/2021/09/watch-out-when-using-pdbbind.\"\n",
     "                )\n",
-    "            if ligand_mol is None:\n",
-    "                raise ValueError(f\"Could not load PDBBind 2020 ligand from {ligand_filepath}.\")\n",
-    "        pc.load_protein_from_pdb(temp_protein_filepath)\n",
-    "        pc.load_ligands_from_mols([ligand_mol])\n",
-    "        pdbbind_protein_ligand_interaction_dfs.append(pc.calculate_interactions())\n",
+    "                try:\n",
+    "                    ligand_mol = Chem.MolFromMolFile(ligand_filepath.replace(\".mol2\", \".sdf\"))\n",
+    "                except Exception as e:\n",
+    "                    ligand_mol = Chem.MolFromMolFile(\n",
+    "                        ligand_filepath.replace(\".mol2\", \".sdf\"), sanitize=False\n",
+    "                    )\n",
+    "                try:\n",
+    "                    Chem.rdmolops.AssignAtomChiralTagsFromStructure(ligand_mol)\n",
+    "                except Exception as e:\n",
+    "                    print(\n",
+    "                        f\"Could not assign chirality tags to the atoms in the PDBBind ligand molecule from {ligand_filepath}.\"\n",
+    "                    )\n",
+    "                if ligand_mol is None:\n",
+    "                    raise ValueError(f\"Could not load PDBBind 2020 ligand from {ligand_filepath}.\")\n",
+    "            pc.load_protein_from_pdb(temp_protein_filepath)\n",
+    "            pc.load_ligands_from_mols([ligand_mol])\n",
+    "            pdbbind_protein_ligand_interaction_dfs.append(pc.calculate_interactions())\n",
+    "        except Exception as e:\n",
+    "            print(\n",
+    "                f\"Error processing PDBBind filepaths {temp_protein_filepath} and {ligand_filepath} due to: {e}. Skipping...\"\n",
+    "            )\n",
+    "            continue\n",
     "\n",
     "        # NOTE: we iteratively save the interaction dataframes to an HDF5 file\n",
     "        with pd.HDFStore(\n",
@@ -397,7 +403,7 @@
     "            dg_protein_ligand_interaction_dfs.append(pc.calculate_interactions())\n",
     "        except Exception as e:\n",
     "            print(\n",
-    "                f\"Error processing Dockgen filepath pari {temp_protein_filepath} and {ligand_filepath} due to: {e}. Skipping...\"\n",
+    "                f\"Error processing Dockgen filepaths {temp_protein_filepath} and {ligand_filepath} due to: {e}. Skipping...\"\n",
     "            )\n",
     "            continue\n",
     "\n",