-
Notifications
You must be signed in to change notification settings - Fork 0
/
predict.py
141 lines (114 loc) · 4.36 KB
/
predict.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
import argparse
import configparser
import os
import time
import pandas as pd
import rdkit.Chem as Chem
import rdkit.Chem.Draw as Draw
import torch
from profis.pred.pred import predict, filter_dataframe
from profis.utils.modelinit import initialize_model
def main(config_path):
"""
Generates structure predictions for the latent embeddings of molecular fingerprints.
Args:
config_path: Path to the config file.
"""
start_time = time.time()
# get config
config = configparser.ConfigParser(allow_no_value=True)
config.read(config_path)
file_path = config["RUN"]["data_path"]
model_path = config["RUN"]["model_path"]
use_cuda = config["RUN"].getboolean("use_cuda")
clf_data_path = config["RUN"]["clf_data_path"]
verbosity = int(config["RUN"]["verbosity"])
n_trials = int(config["RUN"]["n_trials"])
if not os.path.exists(file_path):
raise FileNotFoundError(f"Data file {file_path} not found")
if not os.path.exists(model_path):
raise FileNotFoundError(f"Model file {model_path} not found")
if clf_data_path and not os.path.exists(clf_data_path):
raise FileNotFoundError(f"Classifier train dataset {clf_data_path} not found")
device = torch.device("cuda" if torch.cuda.is_available() and use_cuda else "cpu")
print(f"Using {device} device") if verbosity > 0 else None
# get file name
dirname = os.path.dirname(file_path)
timestamp = time.strftime("%Y%m%d-%H%M%S")
model_config_path = model_path.replace(
model_path.split("/")[-1], "hyperparameters.ini"
)
model_config = configparser.ConfigParser(allow_no_value=True)
if not os.path.exists(model_config_path):
raise ValueError(f"Model config file {model_config_path} not found")
model_config.read(model_config_path)
out_encoding = model_config["RUN"]["out_encoding"]
# load model
model = initialize_model(config_path=model_config_path, device=device)
model.load_state_dict(torch.load(model_path, map_location=device))
print(f"Loaded model from {model_path}") if verbosity > 1 else None
# load data
if file_path.endswith(".csv"):
query_df = pd.read_csv(file_path)
elif file_path.endswith(".parquet"):
query_df = pd.read_parquet(file_path)
else:
raise ValueError("Data file format not supported (must be .csv or .parquet)")
for col in ["smiles", "label", "score", "activity", "norm", "distance_to_model"]:
if col in query_df.columns:
query_df = query_df.drop(columns=[col])
input_vector = query_df.to_numpy()
print(f"Loaded data from {file_path}") if verbosity > 1 else None
# get predictions
print(f"Getting predictions for file {file_path}...") if verbosity > 1 else None
df = predict(
model,
input_vector,
device=device,
format=out_encoding,
batch_size=512,
n_trials=n_trials,
)
# filter dataframe
if len(df) > 0:
df = filter_dataframe(df, config)
else:
print("No valid predictions") if verbosity > 0 else None
# save stats
stats = pd.DataFrame()
stats["mean_qed"] = df["qed"].mean()
# save data as csv
os.mkdir(f"{dirname}/preds_{timestamp}")
with open(f"{dirname}/preds_{timestamp}/config.ini", "w") as configfile:
config.write(configfile)
df.to_csv(f"{dirname}/preds_{timestamp}/predictions.csv", index=False)
(
print(f"Saved data to {dirname}/preds_{timestamp} directory")
if verbosity > 0
else None
)
# dump config
with open(f"{dirname}/preds_{timestamp}/config.ini", "w") as configfile:
config.write(configfile)
# save images
os.mkdir(f"{dirname}/preds_{timestamp}/imgs")
for n, (idx, smiles) in enumerate(zip(df["idx"], df["smiles"])):
mol = Chem.MolFromSmiles(smiles)
Draw.MolToFile(
mol, f"{dirname}/preds_{timestamp}/imgs/{idx}_{n}.png", size=(300, 300)
)
time_elapsed = time.time() - start_time
print(f"{file_path} processed in {(time_elapsed / 60):.2f} minutes")
return
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument(
"--config",
"-c",
type=str,
default="config_files/pred_config.ini",
help="Path to config file",
)
args = parser.parse_args()
config_path = args.config
main(config_path=config_path)