forked from jianglikun/DeepTTC
-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathcross_study_validation.py
198 lines (168 loc) · 8.54 KB
/
cross_study_validation.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 pickle
import sklearn
import numpy as np
import pandas as pd
from pathlib import Path
from scipy.stats import pearsonr, spearmanr
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
sources = ['ccle', 'ctrp', 'gcsi', 'gdsc1', 'gdsc2']
def groupby_src_and_print(df, print_fn=print):
print_fn(df.groupby('SOURCE').agg(
{'CancID': 'nunique', 'DrugID': 'nunique'}).reset_index())
def load_source(src, datadir, use_lincs=True):
pretty_indent = '#' * 10
print('Load source datadir: ', datadir)
print(f'{pretty_indent} {src.upper()} {pretty_indent}')
# Load data
responses = pd.read_csv(
f"{datadir}/rsp_{src}.csv") # Drug response
gene_expression = pd.read_csv(
f"{datadir}/ge_{src}.csv") # Gene expressions
mordred_descriptors = pd.read_csv(
f"{datadir}/mordred_{src}.csv") # Mordred descriptors
morgan_fingerprints = pd.read_csv(
f"{datadir}/ecfp2_{src}.csv") # Morgan fingerprints
smiles = pd.read_csv(f"{datadir}/smiles_{src}.csv") # SMILES
# Use landmark genes
if use_lincs:
with open(f"{datadir}/../landmark_genes") as f:
genes = [str(line.rstrip()) for line in f]
genes = ["ge_" + str(g) for g in genes]
print(len(set(genes).intersection(set(gene_expression.columns[1:]))))
genes = list(set(genes).intersection(set(gene_expression.columns[1:])))
cols = ["CancID"] + genes
gene_expression = gene_expression[cols]
groupby_src_and_print(responses)
print("Unique cell lines with gene expressions",
gene_expression["CancID"].nunique())
print("Unique drugs with Mordred",
mordred_descriptors["DrugID"].nunique())
print("Unique drugs with ECFP2",
morgan_fingerprints["DrugID"].nunique())
print("Unique drugs with SMILES", smiles["DrugID"].nunique())
return gene_expression, mordred_descriptors, morgan_fingerprints, smiles, responses
def score(y_true, y_pred):
scores = {}
scores['r2'] = sklearn.metrics.r2_score(y_true=y_true, y_pred=y_pred)
scores['mean_absolute_error'] = sklearn.metrics.mean_absolute_error(
y_true=y_true, y_pred=y_pred)
scores['spearmanr'] = spearmanr(y_true, y_pred)[0]
scores['pearsonr'] = pearsonr(y_true, y_pred)[0]
print(scores)
return scores
def prepare_dataframe(gene_expression, smiles, responses, model):
gene_expression, drug_data = model.preprocess(
gene_expression, smiles, responses, 'AUC')
drug_data = drug_data.drop(['index'], axis=1)
drug_columns = [
x for x in drug_data.columns if x not in ['CancID', 'DrugID']]
# data = pd.merge(gene_expression, drug_data, on='DrugID', how='inner')
data = pd.merge(gene_expression, drug_data, on='CancID', how='inner')
gene_expression = gene_expression.drop(['CancID'], axis=1)
gene_expression_columns = gene_expression.columns
return data, gene_expression_columns, drug_columns
def run_cross_study_analysis(model, data_dir, results_dir, n_splits=10, use_lincs=True):
for src in sources:
datadir = f"{data_dir}/ml.dfs/July2020/data.{src}"
splitdir = f"{datadir}/splits"
print('Datadir: ', datadir)
gene_expression, _, _, smiles, responses = load_source(
src, datadir, use_lincs)
data, gene_expression_columns, drug_columns = prepare_dataframe(
gene_expression, smiles, responses, model)
print(data)
# -----------------------------------------------
# Train model
# -----------------------------------------------
# Example of training a DRP model with gene expression and Mordred descriptors
print("\nGet the splits.")
for split_id in range(n_splits):
with open(f"{splitdir}/split_{split_id}_tr_id") as f:
train_id = [int(line.rstrip()) for line in f]
with open(f"{splitdir}/split_{split_id}_te_id") as f:
test_id = [int(line.rstrip()) for line in f]
# Train and test data
train_data = data.loc[train_id]
test_data = data.loc[test_id]
test_cancid = test_data['CancID']
test_drugid = test_data['DrugID']
# Val data from tr_data
train_data, validation_data = train_test_split(
train_data, test_size=0.12)
print("Train", train_data.shape)
print("Val ", validation_data.shape)
print("Test ", test_data.shape)
scaler = StandardScaler()
train_data_gene_expression_scaled = pd.DataFrame(
scaler.fit_transform(train_data[gene_expression_columns]))
validation_data_gene_expression_scaled = pd.DataFrame(
scaler.transform(validation_data[gene_expression_columns]))
test_data_gene_expression_scaled = pd.DataFrame(
scaler.transform(test_data[gene_expression_columns]))
# Scale
# Train model
train_data.index = range(train_data.shape[0])
validation_data.index = range(validation_data.shape[0])
test_data.index = range(test_data.shape[0])
model.train(train_drug=train_data[drug_columns], train_rna=train_data_gene_expression_scaled,
val_drug=validation_data[drug_columns], val_rna=validation_data_gene_expression_scaled)
# Predict
# DeepTTC-specific prediction format
_, y_pred, _, _, _, _, _, _, _ = model.predict(
test_data[drug_columns], test_data_gene_expression_scaled)
y_true = test_data['Label']
print('Y_true:')
print(y_true)
# Scores
scores = score(y_true, y_pred)
print(f'CancID: {np.shape(test_cancid)}, DrugID: {np.shape(test_drugid)}, y_true: {np.shape(y_true)}, y_pred: {np.shape(y_pred)}, train_data: {np.shape(train_data)}, test_data: {np.shape(test_data)}')
result = {'CancID': test_cancid,
'DrugID': test_drugid,
'y_true': y_true,
'y_pred': y_pred}
result_df = pd.DataFrame()
for key in result:
print(key)
result_df[key] = np.array(result[key])
print('Y_true 2:')
print(np.shape(result_df))
print(np.shape(y_true))
print(result_df)
# result_df = pd.DataFrame.from_dict(result)
result_df.to_csv(
f'{results_dir}/{src}_{src}_split_{split_id}.csv', sep=',', index=None)
pickle.dump(result, open(
f'{results_dir}/predictions_{src}_cv_split_{split_id}.pickle', 'wb'))
# pickle.dump(scores, open(
# f'{results_dir}/scores_{src}_cv_split_{split_id}.pickle', 'wb'))
# Test on unrelated datasets
for test_src in sources:
if test_src == src:
continue
test_datadir = f"{data_dir}/ml.dfs/July2020/data.{test_src}"
test_gene_expression, _, _, test_smiles, test_responses = load_source(
test_src, test_datadir, use_lincs)
test_src_data, test_gene_expression_columns, test_drug_columns = prepare_dataframe(
test_gene_expression, test_smiles, test_responses, model)
# Subsetting to the current set of expressed genes!!!
test_src_data_gene_expression_scaled = pd.DataFrame(
scaler.transform(test_src_data[gene_expression_columns]))
_, y_pred, _, _, _, _, _, _, _ = model.predict(
test_src_data[test_drug_columns], test_src_data_gene_expression_scaled)
y_true = test_src_data['Label']
test_cancid = test_src_data['CancID']
test_drugid = test_src_data['DrugID']
# print(f'{src} on {test_src} y_true: ')
result = {'CancID': test_cancid,
'DrugID': test_drugid,
'y_true': y_true,
'y_pred': y_pred}
result_df = pd.DataFrame.from_dict(result)
result_df.to_csv(
f'{results_dir}/{src}_{test_src}_split_{split_id}.csv', sep=',', index=None)
scores = score(y_pred, y_true)
# pickle.dump(result, open(
# f'{results_dir}/predictions_{src}_split_{split_id}_on_{test_src}.pickle', 'wb'))
pickle.dump(scores, open(
f'{results_dir}/scores_{src}_split_{split_id}_on_{test_src}.pickle', 'wb'))