forked from bel2scm/bel2scm
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtest_bel2scm.py
290 lines (249 loc) · 12.2 KB
/
test_bel2scm.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
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
import unittest
import os
from bel2scm.neurips_bel2scm.scm import SCM
from bel2scm.neurips_bel2scm.utils import json_load
from bel2scm.neurips_bel2scm.utils import save_scm_object
from bel2scm.neurips_bel2scm.utils import load_scm_object
import torch
import pandas as pd
import time
import numpy as np
from torch import tensor
HERE = os.path.abspath(os.path.dirname(__file__))
DATA = os.path.join(HERE, 'data')
class TestSCM(unittest.TestCase):
def test_mapk_erk_samples(self):
bel_file_path = os.path.join(DATA, "mapk.json")
config_file_path = os.path.join(DATA, "COVID-19-config.json")
data_file_path = os.path.join(DATA, "mapk3000.csv")
# output_pickle_object_file = os.path.join(HERE, "../mapk_scm.pkl"
scm = SCM(bel_file_path, config_file_path, data_file_path)
exogenous_noise = scm.exogenous_dist_dict
samples = [scm.model(exogenous_noise) for i in range(3000)]
print(samples)
# save_scm_object(output_pickle_object_file, scm)
def test_generate_mapk_samples(self):
scm = load_scm_object("../../../mapk_scm.pkl")
exogenous_noise = scm.exogenous_dist_dict
samples = [scm.model(exogenous_noise) for i in range(1000)]
# [TODO] Compare the mean of each variable with data itself.
self.assertTrue(True)
def test_igf(self):
torch.manual_seed(101)
bel_file_path = os.path.join(DATA, "igf.json")
config_file_path = os.path.join(DATA, "COVID-19-config.json")
data_file_path = os.path.join(DATA, "observational_igf.csv")
# output_pickle_object_file = "../../igf_scm.pkl"
scm = SCM(bel_file_path, config_file_path, data_file_path)
# save_scm_object(output_pickle_object_file, scm)
exogenous_noise = scm.exogenous_dist_dict
samples = [scm.model(exogenous_noise) for i in range(5000)]
df = pd.DataFrame(samples)
for col in df.columns:
for i in range(len(df)):
if torch.is_tensor(df[col][i]):
df[col][i] = df[col][i].item()
df.to_csv(os.path.join(DATA, "bel2scm_samples_igf.csv"))
def test_igf_intervention_on_ras(self):
torch.manual_seed(101)
bel_file_path = os.path.join(DATA, "igf.json")
config_file_path = os.path.join(DATA, "COVID-19-config.json")
data_file_path = os.path.join(DATA, "observational_igf.csv")
output_pickle_object_file = "../../../igf_scm.pkl"
scm = SCM(bel_file_path, config_file_path, data_file_path)
exogenous_noise = scm.exogenous_dist_dict
condition_data = scm.model(exogenous_noise)
print(condition_data)
# target = "a(p(Erk))"
intervention_data = {
"a(p(Ras))": 30.0
}
do_model = scm.intervention(intervention_data)
samples = [do_model(exogenous_noise) for _ in range(5000)]
df = pd.DataFrame(samples)
for col in df.columns:
for i in range(len(df)):
if torch.is_tensor(df[col][i]):
df[col][i] = df[col][i].item()
df2 = pd.read_csv(os.path.join(DATA, "bel2scm_samples_igf.csv"))
erk_diff = df["a(p(Erk))"] - df2["a(p(Erk))"]
erk_diff.to_csv(os.path.join(DATA, "erk_do_ras_30_minus_erk.csv"))
df.to_csv(os.path.join(DATA, "intervention_samples_igf.csv"))
def test_igf_intervention_on_mek(self):
torch.manual_seed(23)
bel_file_path = os.path.join(DATA, "igf.json")
config_file_path = os.path.join(DATA, "COVID-19-config.json")
data_file_path = os.path.join(DATA, "observational_igf.csv")
output_pickle_object_file = "../../../igf_scm.pkl"
scm = SCM(bel_file_path, config_file_path, data_file_path)
exogenous_noise = scm.exogenous_dist_dict
condition_data = scm.model(exogenous_noise)
print(condition_data)
# target = "a(p(Erk))"
intervention_data = {
"a(p(Mek))": 40.0
}
do_model = scm.intervention(intervention_data)
samples = [do_model(exogenous_noise) for _ in range(5000)]
df = pd.DataFrame(samples)
for col in df.columns:
for i in range(len(df)):
if torch.is_tensor(df[col][i]):
df[col][i] = df[col][i].item()
df2 = pd.read_csv(os.path.join(DATA, "bel2scm_samples_igf.csv"))
erk_diff = df["a(p(Erk))"] - df2["a(p(Erk))"]
erk_diff.to_csv(os.path.join(DATA, "erk_do_mek_40_minus_erk.csv"))
df.to_csv(os.path.join(DATA, "intervention_mek_40_samples_igf.csv"))
def test_error_with_sde(self):
df_bel = pd.read_csv(os.path.join(DATA, "intervention_samples_igf.csv"))
df_sde = pd.read_csv(os.path.join(DATA, "intervention_igf.csv"))
errors = {}
for col in range(len(df_sde.columns)):
errors[df_sde.columns[col]] = df_sde[df_sde.columns[col]].mean() - df_bel[df_bel.columns[col]].mean()
print(errors)
def test_covid(self):
torch.manual_seed(23)
bel_file_path = os.path.join(DATA, "covid_input.json")
config_file_path = os.path.join(DATA, "COVID-19-config.json")
data_file_path = os.path.join(DATA, "observational_samples_from_sigmoid_known_parameters.csv")
output_pickle_object_file = "../../../covid_scm.pkl"
scm = SCM(bel_file_path, config_file_path, data_file_path)
exogenous_noise = scm.exogenous_dist_dict
samples = [scm.model(exogenous_noise) for i in range(5000)]
df = pd.DataFrame(samples)
for col in df.columns:
for i in range(len(df)):
if torch.is_tensor(df[col][i]):
df[col][i] = df[col][i].item()
df.to_csv(os.path.join(DATA, "bel2scm_samples_covid.csv"))
save_scm_object(output_pickle_object_file, scm)
# Add loading and saving from pkl to utils
def test_covid_direct_simulation_causal_effect(self):
torch.manual_seed(23)
time1 = time.time()
bel_file_path = os.path.join(DATA, "covid_input.json")
config_file_path = os.path.join(DATA, "COVID-19-config.json")
data_file_path = os.path.join(DATA, "observational_samples_from_sigmoid_known_parameters.csv")
scm = SCM(bel_file_path, config_file_path, data_file_path)
exogenous_noise = scm.exogenous_dist_dict
condition_data = scm.model(exogenous_noise)
intervention_data = {
"a(TOCI)": 0.0
}
do_model = scm.intervention(intervention_data)
samples = [do_model(exogenous_noise) for _ in range(5000)]
df = pd.DataFrame(samples)
for col in df.columns:
for i in range(len(df)):
if torch.is_tensor(df[col][i]):
df[col][i] = df[col][i].item()
df2 = pd.read_csv(os.path.join(DATA, "bel2scm_samples_covid.csv"))
cytokine_diff = df["a(cytokine)"] - df2["a(cytokine)"]
cytokine_diff.to_csv(os.path.join(DATA, "cytokine_do_toci_0_minus_cytokine.csv"))
df.to_csv(os.path.join(DATA, "intervention_toci_0_samples_covid.csv"))
def test_covid_noisy_model_samples(self):
torch.manual_seed(23)
time1 = time.time()
bel_file_path = os.path.join(DATA, "covid_input.json")
config_file_path = os.path.join(DATA, "COVID-19-config.json")
data_file_path = os.path.join(DATA, "covid_noisy_reparameterized_data.csv")
scm = SCM(bel_file_path, config_file_path, data_file_path)
exogenous_noise = scm.exogenous_dist_dict
samples = torch.tensor([list(scm.model(exogenous_noise).values()) for _ in range(5500)]).detach().numpy()
np.savetxt(os.path.join(DATA, "covid_bel2scm_noisy_reparameterized_samples.csv"), samples, delimiter=',')
def test_covid_toci0_mm_causal_effect(self):
torch.manual_seed(23)
time1 = time.time()
bel_file_path = os.path.join(DATA, "covid_input.json")
config_file_path = os.path.join(DATA, "COVID-19-config.json")
data_file_path = os.path.join(DATA, "observational_samples_from_sigmoid_known_parameters.csv")
scm = SCM(bel_file_path, config_file_path, data_file_path)
condition_data = {
# 'a(SARS_COV2)': tensor(67.35032),
# 'a(PRR)': tensor(89.7037),
# 'a(ACE2)': tensor(29.747593),
# 'a(AngII)': tensor(68.251114),
# 'a(AGTR1)': tensor(90.96106999999999),
# 'a(ADAM17)': tensor(86.84893000000001),
# 'a(TOCI)': tensor(40.76684),
# 'a(TNF)': tensor(76.85005),
# 'a(sIL_6_alpha)': tensor(87.99491),
# 'a(EGF)': tensor(84.55391),
# 'a(EGFR)': tensor(79.94534),
# 'a(IL6_STAT3)': tensor(83.39896),
# 'a(NF_xB)': tensor(82.79433399999999),
# 'a(IL6_AMP)': tensor(81.38015),
# 'a(cytokine)': tensor(80.21895)
'a(SARS_COV2)': tensor(61.631156999999995),
'a(PRR)': tensor(87.76389),
'a(ACE2)': tensor(39.719845),
'a(AngII)': tensor(59.212959999999995),
'a(AGTR1)': tensor(84.39899399999999),
'a(ADAM17)': tensor(85.84442),
'a(TOCI)': tensor(67.33063),
'a(TNF)': tensor(77.83915),
'a(sIL_6_alpha)': tensor(57.584044999999996),
'a(EGF)': tensor(86.26822),
'a(EGFR)': tensor(81.4849),
'a(IL6_STAT3)': tensor(69.57323000000001),
'a(NF_xB)': tensor(83.75941),
'a(IL6_AMP)': tensor(77.52906),
'a(cytokine)': tensor(79.07555)
}
target = "a(cytokine)"
intervention_data = {
"a(TOCI)": 0.0
}
causal_effects1, counterfactual_samples1 = scm.counterfactual_inference(condition_data, intervention_data,
target, True)
print("time required for causal effects", time.time() - time1)
samples_df = pd.DataFrame(causal_effects1)
samples_df.to_csv(os.path.join(DATA, "causal_effect_MM_bel2scm.csv"), index=False)
def test_covid_toci0_bel2scm_causal_effect(self):
torch.manual_seed(23)
bel_file_path = os.path.join(DATA, "covid_input.json")
config_file_path = os.path.join(DATA, "COVID-19-config.json")
data_file_path = os.path.join(DATA, "observational_samples_from_sigmoid_known_parameters.csv")
scm = SCM(bel_file_path, config_file_path, data_file_path)
condition_data = {
# 'a(SARS_COV2)': tensor(67.35032),
# 'a(PRR)': tensor(89.7037),
# 'a(ACE2)': tensor(29.747593),
# 'a(AngII)': tensor(68.251114),
# 'a(AGTR1)': tensor(90.96106999999999),
# 'a(ADAM17)': tensor(86.84893000000001),
# 'a(TOCI)': tensor(40.76684),
# 'a(TNF)': tensor(76.85005),
# 'a(sIL_6_alpha)': tensor(87.99491),
# 'a(EGF)': tensor(84.55391),
# 'a(EGFR)': tensor(79.94534),
# 'a(IL6_STAT3)': tensor(83.39896),
# 'a(NF_xB)': tensor(82.79433399999999),
# 'a(IL6_AMP)': tensor(81.38015),
# 'a(cytokine)': tensor(80.21895)
'a(SARS_COV2)': tensor(61.631156999999995),
'a(PRR)': tensor(87.76389),
'a(ACE2)': tensor(39.719845),
'a(AngII)': tensor(59.212959999999995),
'a(AGTR1)': tensor(84.39899399999999),
'a(ADAM17)': tensor(85.84442),
'a(TOCI)': tensor(67.33063),
'a(TNF)': tensor(77.83915),
'a(sIL_6_alpha)': tensor(57.584044999999996),
'a(EGF)': tensor(86.26822),
'a(EGFR)': tensor(81.4849),
'a(IL6_STAT3)': tensor(69.57323000000001),
'a(NF_xB)': tensor(83.75941),
'a(IL6_AMP)': tensor(77.52906),
'a(cytokine)': tensor(79.07555)
}
target = "a(cytokine)"
intervention_data = {
"a(TOCI)": 0.0
}
causal_effects1, counterfactual_samples1 = scm.counterfactual_inference(condition_data, intervention_data,
target, True)
samples_df = pd.DataFrame(causal_effects1)
samples_df.to_csv(os.path.join(DATA, "causal_effect_bel2scm_hardcoded_sigmoid_moderately_ill.csv"), index=False)
if __name__ == '__main__':
unittest.main()