-
Notifications
You must be signed in to change notification settings - Fork 0
/
load_dataset.py
204 lines (172 loc) · 9.74 KB
/
load_dataset.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
import os.path
import copy
import pandas as pd
import numpy as np
from rdkit import RDLogger
from rdkit import Chem
from rdkit.Chem import SaltRemover
from molvs import Standardizer
from rdkit.Chem import AllChem
from rdkit import DataStructs
def load_data(name_file, dir_file):
file_in = dir_file + '/' + name_file
df = pd.read_csv(file_in, delimiter=',')
print('\nFile: ', name_file, '\n - ORIGINAL DATA - \n\t Shape: ', df.shape)
print('\t Columns: ', df.columns.tolist())
print('\t Number of cell lines: ', len(df.CELL.unique()))
print('\t Unique NSC: ', len(df.NSC.unique()))
print('\t Number of unique NSC - cell line pairs: ', df.groupby(['NSC', 'CELL']).size().shape)
return df
def load_smiles_file(name_file, dir_file):
file_in = dir_file + '/' + name_file
df_smile = pd.read_csv(file_in, sep='\t', header=None)
df_smile.columns = ['SMILE', 'NSC']
df_smile = df_smile.iloc[:, [1, 0]]
return df_smile
def get_mean(df):
key_info = df["NSC"].map(str) + '_' + df['CELL']
df.insert(0, 'ID', key_info)
df_aux = df.loc[:, ['NLOGGI50', 'ID']]
df_mean = df_aux.groupby('ID', as_index=False).mean()
df_mean.columns = ['ID', 'NLOGGI50_N']
return df_mean
def get_mfp(df_smiles):
mol_list = df_smiles.SMILE.unique()
fingerprints, fingerprints1 = [], []
smiles_list = []
std_smiles = []
error_smiles = []
RDLogger.DisableLog('rdApp.*')
for m in mol_list:
try:
m1 = copy.copy(m)
m = Chem.MolFromSmiles(m)
remover = SaltRemover.SaltRemover() # remove salt
m = remover.StripMol(m)
s = Standardizer() # standardize molecule
m = s.standardize(m)
fp1 = AllChem.GetMorganFingerprintAsBitVect(m, 2, nBits=1024)
fingerprints1.append(fp1)
fingerprints.append(list(fp1))
smiles_list.append(m1)
std_smiles.append(Chem.MolToSmiles(m))
except:
error_smiles.append(m1)
col_name = ['MFP_' + str(i) for i in range(1024)]
df_mfp_aux = pd.DataFrame(data=fingerprints, columns=col_name)
df_std_smile = pd.DataFrame(data=std_smiles, columns=['STD_SMILE'])
df_smile_valid = pd.DataFrame(data=smiles_list, columns=['SMILE'])
df_mfp = pd.concat([df_smile_valid, df_std_smile, df_mfp_aux], axis=1)
return df_mfp, fingerprints1
def remove_few_data(df):
to_remove = ['M19-MEL ', 'DMS 114 ', 'DLD-1 ', 'KM20L2 ', 'SNB-78 ', 'LXFL 529 ', 'DMS 273 ', 'XF 498 ', 'HOP-18 ',
'RXF-631 ', 'MDA-MB-468 ', 'SN12K1 ', 'P388 ', 'P388/ADR ', 'U-251/H.Fine ', 'A-172/H.Fine ',
'U-87/H.Fine ', 'T98G/H.Fine ', 'LN-229/H.Fine ', 'U118/H.Fine ', 'U-373/H.Fine ', 'SK-BR-3 ',
'MAXF 401 ', 'DB ', 'SW-156 ', 'COLO 741 ', 'SW-1573 ', 'UISO-BCA-1 ', 'COLO 746 ', 'CXF 264L ',
'MEXF 514L ', 'TK-164 ', 'HT ', 'RL ', 'H1299p53RE29 ', 'HT29p53RE22 ', 'MCF7-E6 ', 'RKO Waf1 ',
'RKOp53RE1 ', 'T47D FOS1 ', 'T47D NFkB15 ', 'HCT-116/CMV-2 ', 'HCT-116/E6-1 ', 'HCT-116/P ',
'HCT-116/P21/A ', 'HCT-116/P21/B ', 'T47D ERE4 ', 'HCT-116/CMV-1 ', 'HCT-116/PV ', 'HCT-116/P21/C ',
'TSU-PRI ', 'HCT-116/E6-2 ', 'JCA-1 ', 'ND-1 ', 'CACO-2 ', 'ES-2 ', 'MCF7/ATCC ', 'MLI-059 ',
'RPMI-7951 ', 'RXF 486L ', 'SF-767 ', 'SMS-KCNR ', 'SW 1088 ', 'UABLG22 ', 'UABMEL3 ', 'UOK-57 ',
'ZR-75-1 ', 'A-204 ', 'CALU-1 ', 'MLI-045 ', 'MLI-076 ', 'NB4 ', 'CHA-59 ', 'LXFS 650L ', 'MLI-019 ',
'A-C/EBP 3 ', 'A-CREB 1 ', 'A-CREB 2 ', 'A-FOS 2 ', 'A-FOS 3 ', 'A-JUN 1 ', 'A-JUN 3 ', 'A431 ',
'SW 1783 ', 'ZR-75-30 ', 'Mar-Bel ', 'WI-38 ', 'CCD-19LU ', 'MDA-MB-435S ', 'TE85 ', 'VDSO/CMV-9 ',
'VDSO/P ', 'NYH/ICRF-187-1 ', 'VDSO/CMV-8 ', 'VDSO/E6-18 ', 'VDSO/E6-19 ', 'CHO ', 'CHO/159-1 ', 'NYH ']
df_60_cell_lines = df[~df['CELL'].isin(to_remove)]
x_col = df_60_cell_lines["CELL"].replace({'A549/ATCC ': "A549_ATCC", 'OVCAR-8 ': 'OVCAR-8', 'SW-620 ': 'SW-620',
'U251 ': 'U251', 'SF-295 ': 'SF-295', 'NCI-H23 ': 'NCI-H23',
'KM12 ': 'KM12', 'SN12C ': 'SN12C', 'HCT-116 ': 'HCT-116',
'HCT-15 ': 'HCT-15', 'HT29 ': 'HT29', 'SF-268 ': 'SF-268',
'COLO 205 ': 'COLO_205', 'SNB-19 ': 'SNB-19',
'ACHN ': 'ACHN', 'UACC-257 ': 'UACC-257', 'IGROV1 ': 'IGROV1',
'UO-31 ': 'UO-31', 'SK-MEL-28 ': 'SK-MEL-28', '786-0 ': '786-0',
'M14 ': 'M14', 'MOLT-4 ': 'MOLT-4', 'SK-MEL-5 ': 'SK-MEL-5',
'UACC-62 ': 'UACC-62', 'OVCAR-5 ': 'OVCAR-5', 'OVCAR-3 ': 'OVCAR-3',
'K-562 ': 'K-562', 'NCI-H460 ': 'NCI-H460', 'HOP-62 ': 'HOP-62',
'TK-10 ': 'TK-10', 'NCI-H322M ': 'NCI-H322M', 'OVCAR-4 ': 'OVCAR-4',
'SK-OV-3 ': 'SK-OV-3', 'EKVX ': 'EKVX', 'CCRF-CEM ': 'CCRF-CEM',
'LOX IMVI ': 'LOX_IMVI', 'CAKI-1 ': 'CAKI-1', 'SF-539 ': 'SF-539',
'SK-MEL-2 ': 'SK-MEL-2', 'RPMI-8226 ': 'RPMI-8226',
'NCI-H226 ': 'NCI-H226', 'SNB-75 ': 'SNB-75', 'MALME-3M ': 'MALME-3M',
'NCI-H522 ': 'NCI-H522', 'HL-60(TB) ': 'HL-60(TB)', 'RXF 393 ': 'RXF_393',
'HCC-2998 ': 'HCC-2998', 'HOP-92 ': 'HOP-92', 'A498 ': 'A498',
'SR ': 'SR', 'NCI/ADR-RES ': 'NCI_ADR-RES', 'MCF7 ': 'MCF7',
'MDA-MB-435 ': 'MDA-MB-435', 'PC-3 ': 'PC-3', 'DU-145 ': 'DU-145',
'MDA-MB-231/ATCC ': 'MDA-MB-231_ATCC', 'HS 578T ': 'HS_578T',
'T-47D ': 'T-47D', 'BT-549 ': 'BT-549', 'MDA-N ': 'MDA-N'})
df_60_cell_lines.insert(0, 'Cell line', x_col)
f, c = df_60_cell_lines.shape
print('\n Final data: \n\t Cell lines: ', len(df_60_cell_lines.CELL.unique()),
'\n\t Unique smiles used: ', len(df_60_cell_lines.SMILE.unique()),
'\n\t Unique NSC used: ', len(df_60_cell_lines.NSC.unique()),
'\n\t Number of NSC-CELL pairs: ', f)
return df_60_cell_lines
def closest_distance_similarity(fps):
simis = []
nfps = len(fps)
for i in range(1, nfps):
sims = DataStructs.BulkTanimotoSimilarity(fps[i], fps[:i])
max_sims = np.max(np.array(sims))
simis.append(max_sims)
return simis
def check_output_folder(output_directory):
"""
:param output_directory: folder path
:return: None
"""
if not(os.path.isdir(output_directory)):
os.makedirs(output_directory, exist_ok=True)
return None
def get_outlier_molecules(df_aux):
df_tanimoto = pd.DataFrame(data=df_aux['NSC'])
df_fps, fingerprints1 = get_mfp(df_smiles=df_aux)
tanimoto_sims_vect = closest_distance_similarity(fingerprints1)
tanimoto_sims_vect.insert(0, 1)
df_tanimoto['Tanimoto similarity'] = tanimoto_sims_vect
df_tanimoto['outlier_05'] = np.where(df_tanimoto['Tanimoto similarity'] <= 0.5, True, False)
df_new = df_aux.merge(df_tanimoto, on='NSC')
df_nonoutliers = df_new[df_new.loc[:, 'outlier_05'] == False]
print('\tNon-outlier molecules: ', len(df_nonoutliers))
df_fps_non, fingerprints_non = get_mfp(df_smiles=df_nonoutliers)
df_fps_non.insert(0, 'NSC', df_nonoutliers['NSC'].values)
return df_fps_non, fingerprints_non
def get_sample(df, outliers):
df_aux = df.drop_duplicates(subset=['SMILE'], ignore_index=True)
col_name = ['MFP_' + str(i) for i in range(1024)]
if outliers:
print('\nRemoving outliers molecules...')
df_nonoutliers, fingerprints_non = get_outlier_molecules(df_aux)
nsc_col = df_nonoutliers['NSC']
df_fps_non, fingerprints_non = get_mfp(df_smiles=df_nonoutliers)
df_fps_non.insert(0, 'NSC', nsc_col)
df_sample = df_fps_non.loc[:, col_name]
sample = df_sample.to_numpy()
df_nsc_smile = df_fps_non.loc[:, ['NSC', 'SMILE']]
else:
df_sample = df_aux.loc[:, col_name]
sample = df_sample.to_numpy()
df_nsc_smile = df_aux.loc[:, ['NSC', 'SMILE']]
return df_nsc_smile, sample
def get_data_cluster(name_file_nci60, name_file_smi, dir_file, outliers):
df = load_data(name_file=name_file_nci60, dir_file=dir_file)
print('\nPreprocessing data...')
df_mean = get_mean(df)
# Merge both dataframes and drop duplicates
df_with_mean = df.merge(df_mean, how='left', left_on=['ID'], right_on=['ID'])
df_with_mean.drop_duplicates(['NSC', 'CELL'], inplace=True)
df_with_mean.drop(columns=['NLOGGI50'], inplace=True) # remove the previous pGI50 value
# Read smiles file
df_smi = load_smiles_file(name_file=name_file_smi, dir_file=dir_file)
df_mean_smi = df_with_mean.merge(df_smi, how='left', left_on=['NSC'], right_on=['NSC'])
# Remove instances where there is no SMILES information
df_final = df_mean_smi[df_mean_smi.loc[:, 'SMILE'].notnull()]
# MFP
df_mfp_bits, fp1 = get_mfp(df_smiles=df_final)
df_to_process_aux = df_final.merge(df_mfp_bits, how='left', left_on=['SMILE'], right_on=['SMILE'])
df_to_process = df_to_process_aux[df_to_process_aux.loc[:, 'SMILE'].notnull()]
df_to_process.dropna(inplace=True)
# Drop the cell lines with few data
df_60c = remove_few_data(df_to_process)
# get sample
df_info, sample = get_sample(df_60c, outliers)
return df_info, sample