In [1]:

    

from __future__ import division

import os
import numpy as np
import pandas as pd
from scipy import stats
import matplotlib.pyplot as plt
%matplotlib inline


    

In [30]:

    

from rdkit import Chem
from rdkit.Chem import Draw
from rdkit.Chem.Draw import IPythonConsole
from rdkit.Chem.AtomPairs.Pairs import GetAtomPairFingerprint


    

In [3]:

    

txt_dir = "chembl_source"


    

In [4]:

    

# read all chembl bioactivity records
chembl = pd.read_csv(os.path.join(txt_dir, "inhibitor_2017_06_08.csv"), delimiter="\t")
chembl.shape


    

    




/usr/local/lib/python2.7/dist-packages/IPython/core/interactiveshell.py:2717: DtypeWarning: Columns (1,3,6,7,8,9,11,14,16,19,23,27,31,34,35,38,44,48,50,52,53,54,55,56,57,58) have mixed types. Specify dtype option on import or set low_memory=False.
  interactivity=interactivity, compiler=compiler, result=result)






    
Out[4]:





(1235867, 59)

In [5]:

    

# Remove records has no canonical smiles
m = chembl["CANONICAL_SMILES"].isnull()
chembl = chembl[~m]
chembl.shape


    

    
Out[5]:





(1230260, 59)

In [21]:

    

# save inhibitors' smiles and apfp
smiles = chembl[["CMPD_CHEMBLID", "CANONICAL_SMILES"]].copy()
smiles.drop_duplicates(subset="CMPD_CHEMBLID", inplace=True)
smiles.set_index(keys="CMPD_CHEMBLID", drop=True, inplace=True)
smiles.to_csv(txt_dir + "/inhibitor_smiles.csv")


    

In [31]:

    

def dict_2_str(d):
  keylist = d.keys()
  keylist.sort()
  kv_list = ["{}: {}".format(k, d[k]) for k in keylist] 
  return ", ".join(kv_list)

apfp_file = open(txt_dir + "/inhibitor_apfp.csv", "w")
for id_, row in smiles.iterrows():
    m = Chem.MolFromSmiles(row.values[0])
    if m is None:
        print id_
        continue
    apfps = GetAtomPairFingerprint(Chem.RemoveHs(m)).GetNonzeroElements()
    apfp_file.write("%s\t{%s}\n" % (id_, dict_2_str(apfps)))
apfp_file.close()


    

    




CHEMBL1161633
CHEMBL2097021
CHEMBL471869
CHEMBL1161635
CHEMBL181124
CHEMBL1161637
CHEMBL181880
CHEMBL3593577
CHEMBL450200
CMPD_CHEMBLID
CHEMBL450642
CHEMBL2205792
CHEMBL2205793
CHEMBL490121
CHEMBL523281
CHEMBL463327
CHEMBL522826
CHEMBL2205790
CHEMBL495469
CHEMBL2205791
CHEMBL492602
CHEMBL2205788
CHEMBL2205787
CHEMBL452133
CHEMBL2205785
CHEMBL508580
CHEMBL508803
CHEMBL2205789
CHEMBL2205786
CHEMBL493431
CHEMBL2087763
CHEMBL2087764
CHEMBL2179461
CHEMBL2179458
CHEMBL2179464
CHEMBL2179462
CHEMBL2179459
CHEMBL2179463
CHEMBL1083554
CHEMBL2179460
CHEMBL3327018

In [6]:

    

# calculate some molecules's weight
def molwt(x):
    try:
        value = Chem.Descriptors.MolWt(Chem.MolFromSmiles(x))
    except:
        value = np.nan
    return value

m = chembl["MOLWEIGHT"].isnull()
chembl.loc[m, "MOLWEIGHT"] = chembl.loc[m, "CANONICAL_SMILES"].apply(molwt)


    

In [8]:

    

# remove molecules that has no "MOLWEIGHT"
m = chembl["MOLWEIGHT"].isnull()
chembl = chembl[~m]
chembl.shape


    

    
Out[8]:





(1223639, 59)

In [32]:

    

# pick out inhibitor records
inhibitor = chembl[chembl["STANDARD_TYPE"].isin(["IC50", "Ki", "EC50"])]

# inhibitor records: all IC50, a part of Ki and EC50 with "inhibit" in "DESCRIPTION"
m0 = inhibitor["STANDARD_TYPE"].isin(["IC50"]) 
m1 = inhibitor["STANDARD_TYPE"].isin(["Ki", "EC50"]) 
m2 = inhibitor["DESCRIPTION"].apply(lambda x: "inhibit" in x.lower())
m = m0 | (m1 & m2)

inhibitor = inhibitor[m]
inhibitor.shape


    

    
Out[32]:





(835299, 59)

In [33]:

    

# some records without "STANDARD_VALUE" should be cleared away
m = inhibitor["STANDARD_VALUE"].isnull()
inhibitor = inhibitor[~m]
inhibitor.shape


    

    
Out[33]:





(716442, 59)

In [34]:

    

inhibitor["DATA_VALIDITY_COMMENT"].value_counts()


    

    
Out[34]:





Outside typical range            26411
Potential transcription error      378
Non standard unit for type         370
Manually validated                 163
Name: DATA_VALIDITY_COMMENT, dtype: int64

In [35]:

    

# some records with abnormal data also should be cleared away
#error_comment = ["Outside typical range", "Non standard unit for type", "Potential transcription error"]
error_comment = ["Outside typical range"]
m = inhibitor["DATA_VALIDITY_COMMENT"].isin(error_comment)
inhibitor = inhibitor[~m]
inhibitor.shape


    

    
Out[35]:





(690031, 59)

In [36]:

    

# correct some STANDARD_UNITS
m = inhibitor["STANDARD_UNITS"].isin(["/uM"])
inhibitor.loc[m, "STANDARD_VALUE"] = inhibitor.loc[m, "STANDARD_VALUE"].astype(float).values * 1000
inhibitor.loc[m, "STANDARD_UNITS"] = "nM"

m = inhibitor["STANDARD_UNITS"].isin(["/nM", "ug nM-1", "Ke nM-1"])
inhibitor.loc[m, "STANDARD_UNITS"] = "nM"

m = inhibitor["STANDARD_UNITS"].isin(["ug.mL-1"])
inhibitor.loc[m, "STANDARD_VALUE"] = inhibitor.loc[m, "STANDARD_VALUE"].astype(float) / inhibitor.loc[m, "MOLWEIGHT"].astype(float) * 10**6
inhibitor.loc[m, "STANDARD_UNITS"] = "nM"

m = inhibitor["STANDARD_UNITS"].isin(["nM"])
inhibitor = inhibitor[m]
inhibitor.shape


    

    
Out[36]:





(689725, 59)

In [37]:

    

# remove duplicates
m = inhibitor["POTENTIAL_DUPLICATE"].fillna(0).astype(int) == 0
inhibitor = inhibitor[m]
inhibitor.shape


    

    
Out[37]:





(662788, 59)

In [39]:

    

inhibitor.to_csv(txt_dir + "/inhibitor_clean_2017_06_08.csv", index=False)


    

In [ ]:

    




    

In [ ]:

    

# judge a record's clf label
def is_pos(row):
  r = row["RELATION"]
  v = np.float32(row["STANDARD_VALUE"])
  if r == "<" or r == "<=":
    return 1 if v <= 10000 else np.nan
  elif r == ">" or r == ">=":
    return -1 if v >= 10000 else np.nan
  elif r == "=":
    return 1 if v <= 10000 else -1
  else:
    return np.nan


    

In [89]:

    

inhibitor["CLF_LABEL"] = inhibitor.apply(is_pos, axis=1)
inhibitor = inhibitor[~inhibitor["CLF_LABEL"].isnull()]
inhibitor.loc[:, "YEAR"] = inhibitor.loc[:, "YEAR"].astype(float)


    

In [131]:

    

# group
grouped = inhibitor.groupby(by=["TARGET_CHEMBLID", "PREF_NAME", "CMPD_CHEMBLID"], as_index=False)
# judge one molecule's label by the average label
clf_label = grouped[["CLF_LABEL", "YEAR"]].mean()
clf_label


    

    
Out[131]:






  

    

      

      
TARGET_CHEMBLID
      
PREF_NAME
      
CMPD_CHEMBLID
      
CLF_LABEL
      
YEAR
    
  
  

    

      
0
      
CHEMBL1075092
      
Glycine receptor subunit alpha-3
      
CHEMBL1092618
      
-1.0
      
2010.0
    
    

      
1
      
CHEMBL1075092
      
Glycine receptor subunit alpha-3
      
CHEMBL1092619
      
-1.0
      
2010.0
    
    

      
2
      
CHEMBL1075092
      
Glycine receptor subunit alpha-3
      
CHEMBL1093582
      
-1.0
      
2010.0
    
    

      
3
      
CHEMBL1075092
      
Glycine receptor subunit alpha-3
      
CHEMBL1093848
      
-1.0
      
2010.0
    
    

      
4
      
CHEMBL1075092
      
Glycine receptor subunit alpha-3
      
CHEMBL2398350
      
-1.0
      
2013.0
    
    

      
5
      
CHEMBL1075092
      
Glycine receptor subunit alpha-3
      
CHEMBL2398352
      
-1.0
      
2013.0
    
    

      
6
      
CHEMBL1075092
      
Glycine receptor subunit alpha-3
      
CHEMBL464651
      
1.0
      
2010.0
    
    

      
7
      
CHEMBL1075097
      
Arginase-1
      
CHEMBL1099169
      
1.0
      
2010.0
    
    

      
8
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3714817
      
1.0
      
NaN
    
    

      
9
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3714879
      
1.0
      
NaN
    
    

      
10
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3714885
      
1.0
      
NaN
    
    

      
11
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3714909
      
1.0
      
NaN
    
    

      
12
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3714960
      
1.0
      
NaN
    
    

      
13
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3714970
      
1.0
      
NaN
    
    

      
14
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3715004
      
1.0
      
NaN
    
    

      
15
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3715017
      
1.0
      
NaN
    
    

      
16
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3715077
      
1.0
      
NaN
    
    

      
17
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3715155
      
1.0
      
NaN
    
    

      
18
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3715174
      
1.0
      
NaN
    
    

      
19
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3715218
      
1.0
      
NaN
    
    

      
20
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3715359
      
1.0
      
NaN
    
    

      
21
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3715361
      
1.0
      
NaN
    
    

      
22
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3715375
      
1.0
      
NaN
    
    

      
23
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3715394
      
1.0
      
NaN
    
    

      
24
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3715478
      
1.0
      
NaN
    
    

      
25
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3715536
      
1.0
      
NaN
    
    

      
26
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3715558
      
1.0
      
NaN
    
    

      
27
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3715572
      
1.0
      
NaN
    
    

      
28
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3715577
      
1.0
      
NaN
    
    

      
29
      
CHEMBL1075101
      
G-protein coupled receptor 81
      
CHEMBL3715599
      
1.0
      
NaN
    
    

      
...
      
...
      
...
      
...
      
...
      
...
    
    

      
542203
      
CHEMBL6175
      
Lysine-specific demethylase 4C
      
CHEMBL3786862
      
1.0
      
2016.0
    
    

      
542204
      
CHEMBL6175
      
Lysine-specific demethylase 4C
      
CHEMBL3786952
      
1.0
      
2016.0
    
    

      
542205
      
CHEMBL6175
      
Lysine-specific demethylase 4C
      
CHEMBL3786963
      
1.0
      
2016.0
    
    

      
542206
      
CHEMBL6175
      
Lysine-specific demethylase 4C
      
CHEMBL3787020
      
1.0
      
2016.0
    
    

      
542207
      
CHEMBL6175
      
Lysine-specific demethylase 4C
      
CHEMBL3787044
      
1.0
      
2016.0
    
    

      
542208
      
CHEMBL6175
      
Lysine-specific demethylase 4C
      
CHEMBL3787133
      
1.0
      
2016.0
    
    

      
542209
      
CHEMBL6175
      
Lysine-specific demethylase 4C
      
CHEMBL3787193
      
1.0
      
2016.0
    
    

      
542210
      
CHEMBL6175
      
Lysine-specific demethylase 4C
      
CHEMBL3787438
      
1.0
      
2016.0
    
    

      
542211
      
CHEMBL6175
      
Lysine-specific demethylase 4C
      
CHEMBL3787516
      
1.0
      
2016.0
    
    

      
542212
      
CHEMBL6175
      
Lysine-specific demethylase 4C
      
CHEMBL3787534
      
1.0
      
2016.0
    
    

      
542213
      
CHEMBL6175
      
Lysine-specific demethylase 4C
      
CHEMBL3787548
      
1.0
      
2016.0
    
    

      
542214
      
CHEMBL6175
      
Lysine-specific demethylase 4C
      
CHEMBL3787556
      
1.0
      
2016.0
    
    

      
542215
      
CHEMBL6175
      
Lysine-specific demethylase 4C
      
CHEMBL3787664
      
1.0
      
2016.0
    
    

      
542216
      
CHEMBL6175
      
Lysine-specific demethylase 4C
      
CHEMBL3787669
      
1.0
      
2016.0
    
    

      
542217
      
CHEMBL6177
      
NAD kinase
      
CHEMBL233434
      
-1.0
      
2008.0
    
    

      
542218
      
CHEMBL6177
      
NAD kinase
      
CHEMBL538665
      
-1.0
      
2009.0
    
    

      
542219
      
CHEMBL6177
      
NAD kinase
      
CHEMBL560315
      
1.0
      
2009.0
    
    

      
542220
      
CHEMBL6177
      
NAD kinase
      
CHEMBL561654
      
-1.0
      
2009.0
    
    

      
542221
      
CHEMBL6177
      
NAD kinase
      
CHEMBL562056
      
1.0
      
2009.0
    
    

      
542222
      
CHEMBL6186
      
Serine/threonine-protein kinase Sgk3
      
CHEMBL2333365
      
-1.0
      
2013.0
    
    

      
542223
      
CHEMBL6186
      
Serine/threonine-protein kinase Sgk3
      
CHEMBL3092460
      
-1.0
      
2015.0
    
    

      
542224
      
CHEMBL6186
      
Serine/threonine-protein kinase Sgk3
      
CHEMBL3092468
      
1.0
      
2015.0
    
    

      
542225
      
CHEMBL6186
      
Serine/threonine-protein kinase Sgk3
      
CHEMBL3745885
      
-1.0
      
2016.0
    
    

      
542226
      
CHEMBL6195
      
Ubiquitin carboxyl-terminal hydrolase isozyme L3
      
CHEMBL1190585
      
-1.0
      
2007.0
    
    

      
542227
      
CHEMBL6195
      
Ubiquitin carboxyl-terminal hydrolase isozyme L3
      
CHEMBL1241028
      
1.0
      
2007.0
    
    

      
542228
      
CHEMBL6195
      
Ubiquitin carboxyl-terminal hydrolase isozyme L3
      
CHEMBL1241672
      
-1.0
      
2007.0
    
    

      
542229
      
CHEMBL6195
      
Ubiquitin carboxyl-terminal hydrolase isozyme L3
      
CHEMBL1241673
      
-1.0
      
2007.0
    
    

      
542230
      
CHEMBL6195
      
Ubiquitin carboxyl-terminal hydrolase isozyme L3
      
CHEMBL1241765
      
-1.0
      
2007.0
    
    

      
542231
      
CHEMBL6195
      
Ubiquitin carboxyl-terminal hydrolase isozyme L3
      
CHEMBL1241766
      
-1.0
      
2007.0
    
    

      
542232
      
CHEMBL6195
      
Ubiquitin carboxyl-terminal hydrolase isozyme L3
      
CHEMBL590
      
-1.0
      
2007.0
    
  

542233 rows × 5 columns

In [105]:

    

clf_label.to_csv(txt_dir + "/inhibitor_clf_label.csv")


    

In [ ]:

    




    

In [144]:

    

cancer_approved_target = ["CHEMBL279", "CHEMBL203", "CHEMBL333", "CHEMBL325", "CHEMBL267", "CHEMBL2842"]
cancer_clinical_target = ["CHEMBL340", "CHEMBL4005", "CHEMBL332"]


    

In [158]:

    

for target in cancer_approved_target + cancer_clinical_target:
    df = clf_label[clf_label["TARGET_CHEMBLID"] == target]
    df.to_csv(txt_dir + "/%s_clf_label.csv" % target, index=False)


    

In [ ]: