-
Notifications
You must be signed in to change notification settings - Fork 1
/
EquiPNAS.py
152 lines (130 loc) · 5.64 KB
/
EquiPNAS.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
#!/usr/bin/python
# EquiPNAS: Improved protein-nucleic binding site prediction
# using pretrained protein language model and equivariant deep graph learning
#
# Copyright (C) Bhattacharya Laboratory 2023
#
# EquiPNAS is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# EquiPNAS is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with ProXiGram. If not, see <http://www.gnu.org/licenses/>.
#
############################################################################
from egnn_clean import *
import argparse
import os
import sys
import warnings
warnings.simplefilter(action='ignore', category=FutureWarning)
from dgl.dataloading import GraphDataLoader
import dgl
import math
import numpy as np
import torch
from Dataloader import buildGraph
from torch import nn, optim
from torch.nn import functional as F
from torch.utils.data import DataLoader
def to_np(x):
return x.cpu().detach().numpy()
def test_epoch(epoch, model, dataloader, PARS):
model.eval()
rloss = 0
for i, (data_feats) in enumerate(dataloader):
(tgt_name, nodeFeats, xyz_feats, edges, edge_att) = data_feats
tgt_name = tgt_name[0]
#print(tgt_name)
print('running ' + tgt_name + ' ...')
n_nodes = len(nodeFeats[0])
n_e = len(edges[0])
nodeFeats = nodeFeats.to(PARS.device)
xyz_feats = xyz_feats.to(PARS.device)
edges[0] = edges[0].to(PARS.device)
edges[1] = edges[1].to(PARS.device)
edge_att = edge_att.to(PARS.device)
nodeFeats = nodeFeats.squeeze()
xyz_feats = xyz_feats.squeeze()
edges[0] = edges[0].squeeze()
edges[1] = edges[1].squeeze()
edge_att = edge_att.squeeze()
edge_att = edge_att.unsqueeze(dim=1)
pred, xyz = model(nodeFeats, xyz_feats, edges, edge_att)
pred = torch.nn.Sigmoid()(pred)
pred = pred.detach().numpy()
#print(pred)
fo = open(PARS.outdir + '/' + tgt_name + '.out', 'w')
for pr in pred:
fo.write(str(pr[0]) + '\n')
fo.close()
print('done!')
def print_usage():
print("\nUsage: EquiPNAS.py [options]\n")
print("Options:")
print(" -h, --help show this help message and exit")
print(" --model_state_dict MODEL_STATE_DICT")
print(" Saved model")
print(" --indir INDIR Path to input data containing distance maps and input features (default 'datasets/DNA_test_129_Preprocessing_using_AlphaFold2/')")
print(" --outdir OUTDIR Prediction output directory")
print(" --num_workers NUM_WORKERS")
print(" Number of workers (default=4)")
def main(PARS):
dataset = buildGraph(PARS.indir)
inference_loader = GraphDataLoader(dataset, batch_size=1, shuffle=False)
# Get Model
model = EGNN(in_node_nf=5461, hidden_nf=768, out_node_nf=1, in_edge_nf=1, n_layers=12,
attention=True)
if not PARS.model_state_dict:
print("PARS.model_state_dict file must be set")
model.load_state_dict(torch.load(PARS.model_state_dict))
model.to(PARS.device)
test_epoch(0, model, inference_loader, PARS)
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--model_state_dict', type=str, default=None,
help="Saved model")
parser.add_argument('--indir', type=str, default='datasets/DNA_test_129_Preprocessing_using_AlphaFold2/',
help="Path to input data containing distance maps and input features (default 'datasets/DNA_test_129_Preprocessing_using_AlphaFold2/')")
parser.add_argument('--outdir', type=str, default='',
help="Prediction output directory")
parser.add_argument('--num_workers', type=int, default=4,
help="Number of workers (default=4)")
PARS, _ = parser.parse_known_args()
#basic input check
if not PARS.model_state_dict:
print('Error! Trained model must be provided. Exiting ...')
print_usage()
sys.exit()
if (PARS.outdir == ''):
print('Error! Path to the output directory must be provided. Exiting ...')
print_usage()
sys.exit()
#existance check
if not os.path.exists(PARS.model_state_dict):
print('Error! No such trained model exists. Exiting ...')
print_usage()
sys.exit()
if not os.path.exists(PARS.outdir):
print('Error! No such output directory exists. Exiting ...')
print_usage()
sys.exit()
#header
print("\n********************************************************************************")
print("* EquiPNAS *")
print("* Improved protein-nucleic binding site prediction *")
print("* using pretrained protein language model and equivariant deep graph learning *")
print("* For comments, please email to dbhattacharya@vt.edu *")
print("********************************************************************************\n")
print('Residie-level predictions for each target is being saved at ' + PARS.outdir + '/\n')
seed = 1992
torch.manual_seed(seed)
np.random.seed(seed)
PARS.device = torch.device('cpu')
main(PARS)