fix bugs

README.md

@@ -7,7 +7,7 @@ We recommend to try out our web application for that: https://deepimmuno.herokua
 
 > Plase refer to **DeepImmuno-GAN** if you want to generate immunogenic peptide
 
-Enjoy and don't hesitate to ask me questions (contact at the bottom), I will be responsive!
+Enjoy and don't hesitate to ask me questions (contact at the bottom), I will be responsive! Feel free to raise an issue on github page!
 
 ## Citation
 If you find that tool useful in your research, please consider citing our preprint:
@@ -25,12 +25,16 @@ numpy = 1.18.5
 
 pandas = 1.1.1
 
+```
+Note: This is the enviroment that I used for development and I also tested it. But as long as you use python > 3, tensorflow = 2.3, It should also work.
+```
+
 #### How to use?
 
 If you want to query a single epitope (peptide + HLA), for example you want to query peptide _**HPPLMNVER**_ along with _**HLA-A*0201**_. You need to
 
 ```
-python3 deepimmuno-cnn.py --mode "single" --epitope "HPPLMNVER" --HLA "HLA-A*0201"
+python3 deepimmuno-cnn.py --mode "single" --epitope "HPPLMNVER" --hla "HLA-A*0201"
 ```
 
 If you want to query multiple epitopes, you just need to prepare a csv file like this:
@@ -47,6 +51,10 @@ Then you run:
 python3 deepimmuno-cnn.py --mode "multiple" --intdir "/path/to/above/file" --outdir "/path/to/output/folder"
 ```
 
+**1. Please note, when you specify the output dir, don't include the forward slash at the end, for example, use "~/Desktop" instead "~/Desktop/"**
+
+**2. PLease note, if python3 doesn't work, you can replace python3 to python, it depends your installed python interpreter**
+
 A full help prompt is as below:
 
 ```
@@ -76,6 +84,10 @@ numpy = 1.18.4
 
 pandas = 1.0.5
 
+```
+Note: This is the enviroment that I used for development and I also tested it. But as long as you use python > 3, pytorch = 1.4, It should also work.
+```
+
 #### How to use
 
 Pretty simple, just run like this
@@ -97,6 +109,10 @@ optional arguments:
   --outdir OUTDIR  specifying your output folder
 ```
 
+**1. Please note, when you specify the output dir, don't include the forward slash at the end, for example, use "~/Desktop" instead "~/Desktop/"**
+
+**2. PLease note, if python3 doesn't work, you can replace python3 to python, it depends your installed python interpreter**
+
 ## Contact
 
 Guangyuan(Frank) Li

data/hla2paratopeTable_aligned.txt

@@ -0,0 +1,63 @@
+HLA	pseudo
+HLA-A*0203	------------YYEKVAH---TVDTLYRY-YYTKWEQWYTWY---
+HLA-A*0205	------Y-----YYEKVV---------YDYYRDTKWQWYWY-----
+HLA-A*0206	-----MY-----YYEKVAH---TVDTLYRYHYYTKWVQLYTWY---
+HLA-A*0207	------M-----YYEKVAH---TVDTLYR-HYYTKWVQLYTWY---
+HLA-A*3001	------M-----YYENV-QT-----DTLYIYHYTKWWQLYWY----
+HLA-A*6801	-----MY-----YYRNNAQT---V--DTLYYDYTKWAVQWYTWY--
+HLA-A*9235	------M-----YYEKV-HT-----DTYRYHYYTKWVQLYWY----
+HLA-A*9253	------M-----YYEKHT---------VDTRYYTKWYTWY------
+HLA-B*0602	----YDS------YEKYRQ-------ANKYWYYTKWEQWYWY----
+HLA-B*1557	---------------YQWN-----TLYRYEFYWTKWAAYW------
+HLA-B*1801	------Y----HSYRNITNTYES-NLYLR-YSYTKWVQYTW-----
+HLA-B*3505	-------------YSGFYNDMNH-NSYPG-----------------
+HLA-B*4001	---MYHT-----KYREITNT-ESNY-----RYYSKVQYWY------
+HLA-B*4002	---MYHT----FKYREISNTYES-NLYLSYYYITKWVQLYEWY---
+HLA-B*4044	---MYHT----FKYREISNTYES-NLYLSYYYITKWVQLYEWY---
+HLA-B*4102	----YHT-----KYREIT---ESNL---YLYYYTKWQDYWY-----
+HLA-B*4104	---MYHT-----KYREISNT-ESNL---YLYYYTKWAVQDYTWY--
+HLA-B*4202	------------YYRNIYAQTES-NLYL--YYYTKWAVQDYWY---
+HLA-B*4601	-----------MYYREKY-QTVS-NLY---RYYTKWEQWYLWY---
+HLA-B*5706	------M-----YYENSTNI---------YIYYIKWQLYWY-----
+HLA-B*5712	------M-----YYENMTYN------NYIYDYWTKWQLYWY-----
+HLA-B*8103	-----------YYYRNIYAQTES-NYY---NYYSKVQDYEWY----
+HLA-B*9234	------------MYEKH--T-----DTLYRYYYTKWVQYTWY----
+HLA-C*0102	-----------YHYRES--ATIF-NT-YWHFYWSKSEHQYWY----
+HLA-C*0304	------M-----YREKRQT-----SNLY--RYYYTWEQYWY-----
+HLA-C*0401	------Y-----YREKYRQD----NKLYLRFFYTKWRYWY------
+HLA-C*0517	------M-----YYEKRQT-----NKLYRYNFYTKWERYWY-----
+HLA-C*0602	MYDSYEK------YRQADV-------NKLYWYYTKWQWYWY-----
+HLA-C*0756	-----YD----FSYREKY-QAVS-NLYR--SDYTKLAQLYTWY---
+HLA-C*1604	-----------YYYRNIFNTYES-NLY---RYYTKWQLYLWY----
+HLA-A*0101	------M-----YYQENTHT---NTLYIIYRDYTKWAQRYYRGRY-
+HLA-A*0201	---MYMM-----YYEKVVHHTHTVDTLYRYYYYTKWVQLYYYYY--
+HLA-A*0224	----MYM-----YYEKH-HT----DTLYRYYYYTKWVQLYWY----
+HLA-A*0301	-----MY-----YQENVVAQ------DTYYRDYTKWEQLYTWY---
+HLA-A*0362	------M-----YYENA-QT-----DTLYYRDYTKWEQLYWY----
+HLA-A*1101	------Y-----YYENAQV---TVDTLYYYRDYTKWAAQQYWY---
+HLA-A*2301	------------YYEKVAHT------NIYLFYYTKWVQQYTGY---
+HLA-A*2402	------M-----YYEKKVHT----NIYLMFHYYTKWVQYYTYRY--
+HLA-A*3003	----YSM-----YYENVHTE----NTLYIYEHYTKWRLYTWY----
+HLA-A*9234	------------MYEKV-HT----HDT-YRYYYTKWVLLYWY----
+HLA-B*0702	-----------YYYRNIY-QTES-NLYYY-DYYTKWEQRYEWY---
+HLA-B*0801	---MYDY----YFYRNIFNTDES-NLYLSYNYYTKWVVQDYWYY--
+HLA-B*1402	---YYSY----YSEYNICTNTES-NLYLW-YFYTKWELYTW-----
+HLA-B*1501	MYYAMYY----YYREI--NTYES-NLYLRYSYYTKWAEQWYLWY--
+HLA-B*2703	----YHT-----EHREICTE-D---TLY-LYDYTKWVQLYEWY---
+HLA-B*2705	---YYHT----EYREEICTE-DEDTLYL-YYDYTKWVQLAYEWY--
+HLA-B*2709	---MYHT----EYREQICTE-TDDTY---YYHYTKWVVQLYEWY--
+HLA-B*2713	---MYHTEVVRFYREEICAK-DTDYYYHYHDAYTKWVRQLYYECWY
+HLA-B*3501	-----MM----YYRNIFTNTYES-NLYIRYSSYTKWAVQLYWY---
+HLA-B*3508	----MMY----YYRQIITNTYES-NLYIR-YSYTKWQRWYYWY---
+HLA-B*3901	-----YY----YSEYNIC-TTES-NLYLR-YFYTKWVQYYWY----
+HLA-B*4201	------------YYRNIYAQTES-NY-L--YYYTKWAVQDYWY---
+HLA-B*4402	---MYYT-----KYREISNT-ENNTT--YIYDYTKWVVQDYLSRY-
+HLA-B*4403	---MYYT-----KYREITTE---NTY--IRYDYTKWVVQLYLSRY-
+HLA-B*4405	---MYYT-----KYREITTE---NTY--I-YYYTKWVVQDYLSRY-
+HLA-B*5101	------Y-----YYRNINT------YNIYWYYYTKWEQLYLW----
+HLA-B*5301	------Y-----YYRNIFT----NTENIYIRYYTKWVQLYWY----
+HLA-B*5701	------M-----YYENTYNI------YIYDSYWTKWVQLYWY----
+HLA-B*5703	----MYY-----YEENA-ST----NTYNIYYYYTKWVQLYWYY---
+HLA-B*5801	------------YYENSTYE---NIYIRYSAYYTKWVQLYWYY---
+HLA-B*8102	------M----YYYRNIYAQTES-NLYY--NYYSKAVQLYWY----
+HLA-C*1510	-----MY-----YYENMQTD---DNIYIYIYDYTKWVQLYLWY---

deepimmuno-cnn.py

@@ -10,8 +10,6 @@
 from tensorflow.keras import layers
 import numpy as np
 import pandas as pd
-import matplotlib.pyplot as plt
-from sklearn.metrics import auc,precision_recall_curve,roc_curve,confusion_matrix
 import argparse
 import os
 
@@ -138,7 +136,7 @@ def rescue_unknown_hla(hla, dic_inventory):
 
 
 
-def hla_data_aaindex(hla_dic,hla_type,after_pca):    # return numpy array [34,12,1]
+def hla_data_aaindex(hla_dic,hla_type,after_pca,dic_inventory):    # return numpy array [46,12,1]
     try:
         seq = hla_dic[hla_type]
     except KeyError:
@@ -148,7 +146,7 @@ def hla_data_aaindex(hla_dic,hla_type,after_pca):    # return numpy array [34,12
     encode = encode.reshape(encode.shape[0], encode.shape[1], -1)
     return encode
 
-def construct_aaindex(ori,hla_dic,after_pca):
+def construct_aaindex(ori,hla_dic,after_pca,dic_inventory):
     series = []
     for i in range(ori.shape[0]):
         peptide = ori['peptide'].iloc[i]
@@ -157,7 +155,7 @@ def construct_aaindex(ori,hla_dic,after_pca):
 
         encode_pep = peptide_data_aaindex(peptide,after_pca)    # [10,12]
 
-        encode_hla = hla_data_aaindex(hla_dic,hla_type,after_pca)   # [46,12]
+        encode_hla = hla_data_aaindex(hla_dic,hla_type,after_pca,dic_inventory)   # [46,12]
         series.append((encode_pep, encode_hla, immuno))
     return series
 
@@ -167,6 +165,7 @@ def hla_df_to_dic(hla):
         col1 = hla['HLA'].iloc[i]  # HLA allele
         col2 = hla['pseudo'].iloc[i]  # pseudo sequence
         dic[col1] = col2
+    return dic
 
 
 
@@ -217,13 +216,19 @@ def file_process(upload,download):
 def main(args):
     mode = args.mode
     if mode == 'single':
+        print("mode is single")
         epitope = args.epitope
+        print("queried epitope is {}".format(epitope))
         hla= args.hla
+        print("queried epitope is {}".format(hla))
         score = computing_s(epitope,hla)
         print(score)
     elif mode == 'multiple':
+        print("mode is multiple")
         intFile = args.intdir
+        print("input file is {}".format(intFile))
         outFolder = args.outdir
+        print("output will be in {}".format(outFolder))
         file_process(intFile,outFolder)
 
 

deepimmuno-gan.py

@@ -7,9 +7,6 @@
 import torch.nn.functional as F
 import numpy as np
 import pandas as pd
-import matplotlib as mpl
-import matplotlib.pyplot as plt
-from matplotlib.colors import ListedColormap
 import argparse
 import os
 
@@ -126,6 +123,16 @@ def __len__(self):
 
 # auxiliary function during training GAN
 def sample_generator(batch_size):
+    batch_size = 64
+    lr = 0.0001
+    num_epochs = 100
+    seq_len = 10
+    hidden = 128
+    n_chars = 21
+    d_steps = 10
+    device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
+    G = Generator(hidden,seq_len,n_chars,batch_size).to(device)
+    G.load_state_dict(torch.load('./models/wassGAN_G.pth'))
     noise = torch.randn(batch_size, 128).to(device)  # [N, 128]
     generated_data = G(noise)  # [N, seq_len, n_chars]
     return generated_data
@@ -285,6 +292,7 @@ def inverse_transform(hard):  # [N,seq_len]
 def main(args):
     #batch= args.batch
     outdir = args.outdir
+    print("outdir is {}".format(outdir))
     generation = sample_generator(64).detach().cpu().numpy()  # [N,seq_len,n_chars]
     hard = np.argmax(generation, axis=2)  # [N,seq_len]
     pseudo = inverse_transform(hard)