-
Notifications
You must be signed in to change notification settings - Fork 0
/
train.py
172 lines (121 loc) · 5.4 KB
/
train.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
#!/usr/bin/env python
# coding: utf-8
import torch
import torch.nn as nn
from affinity_module.config import get_config
import numpy as np
from dgllife.utils import EarlyStopping
from affinity_module.utils import set_random_seed, load_model
from affinity_module.utils import run_a_train_epoch, run_stat_epoch, run_an_eval_epoch
from affinity_module.utils import Collate
from affinity_module.dataset import VPLGDataset, FoldsOf_VPLGDataset
from affinity_module.protein_graph_loaders import DSSP_loader
from torch.backends import cudnn
cudnn.deterministic = True
cudnn.benchmark = False
args = get_config()
collate = Collate(args)
args['device'] = torch.device("cuda: 0") if torch.cuda.is_available() else torch.device("cpu")
set_random_seed(args['random_seed'])
argv_valFold = args['argv_valFold']
argv_testFold = args['argv_testFold']
cache_dir_prefix = args['cache_dir_prefix']
pdb2graph_translator = DSSP_loader(dssp_files_path = args['dssp_files_path'],
includeAminoacidPhyschemFeatures = False,
cache_dir_prefix = cache_dir_prefix)
best_model_filename = pdb2graph_translator.get_best_model_filename()
_colNames = dict(master_data_table = args['master_data_table'],
pdb_id_col_name="PDBs", smiles_col_name="SMILES", target_col_name="logKi",
foldId_col_name='Fold')
dataset = VPLGDataset(
smiles_to_graph=args['smiles_to_graph'],
smiles_node_featurizer=args['smiles_node_featurizer'],
smiles_edge_featurizer=args['smiles_edge_featurizer'],
**_colNames,
pdb2graph_translator = pdb2graph_translator,
load=False)
args['device'] = torch.device("cuda: 0") if torch.cuda.is_available() else torch.device("cpu")
raw_dataset = VPLGDataset(
smiles_to_graph=args['smiles_to_graph'],
smiles_node_featurizer=args['smiles_node_featurizer'],
smiles_edge_featurizer=args['smiles_edge_featurizer'],
**_colNames,
pdb2graph_translator = pdb2graph_translator,
load=True)
args['fasta_node_feat_size'] = raw_dataset.fasta_graphs[0].ndata['h'].shape[1]
args['fasta_edge_feat_size'] = raw_dataset.fasta_graphs[0].edata['e'].shape[1]
print(args['fasta_node_feat_size'], args['fasta_edge_feat_size'])
print('will save best model to:', best_model_filename)
shfl = True
p = dict(batch_size=args['batch_size'], shuffle=shfl, collate_fn=collate.collate_molgraphs)
mx_nodes = 5000 # filter out large graphs
folds5 = [0,1,2,3,4]
folds5.remove(argv_valFold)
if argv_testFold in folds5:
folds5.remove(argv_testFold)
print('Folds to use: train=%s, val=%s, test=%s' % (str(folds5), str(argv_valFold), str(argv_testFold)) )
#
train_loader = FoldsOf_VPLGDataset(raw_dataset, folds5, max_nodes = mx_nodes).asDataLoader(**p)
val_loader = FoldsOf_VPLGDataset(raw_dataset, [argv_valFold], max_nodes = mx_nodes ).asDataLoader(**p)
test_loader = FoldsOf_VPLGDataset(raw_dataset, [argv_testFold], max_nodes = mx_nodes ).asDataLoader(**p)
print('number of batches: train %d, val %d, test %d' % (len(train_loader), len(val_loader), len(test_loader)))
torch.cuda.empty_cache()
model = load_model(args)
loss_fn = nn.MSELoss(reduction='none')
optimizer = torch.optim.Adam(model.parameters(), lr=args['lr'],
weight_decay=args['weight_decay'])
stopper = EarlyStopping(mode=args['mode'],
patience=args['patience'],
filename=best_model_filename)
if args['load_checkpoint']:
print('Loading checkpoint...')
stopper.load_checkpoint(model)
model.to(args['device'])
for epoch in range(args['num_epochs']):
# Train
run_a_train_epoch(args, epoch, model, train_loader, loss_fn, optimizer)
# Validation and early stop
val_score_ext = run_stat_epoch(args, model, val_loader)
val_score = val_score_ext[ args['metric_name'] ]
test_score = run_an_eval_epoch(args, model, test_loader)
early_stop = stopper.step(val_score, model)
print('epoch {:d}/{:d}, validation {} {:.4f}, test {} {:.4f}, best validation {} {:.4f}'.format(
epoch + 1, args['num_epochs'], args['metric_name'], val_score,
args['metric_name'], test_score,
args['metric_name'], stopper.best_score),
', now R2 = %.4f' % val_score_ext['R2'])
if early_stop:
break
print('-'*80)
stopper.load_checkpoint(model)
print()
all_metrics = {}
for dsName, data_loader in zip(['train', 'val', 'test'], [train_loader, val_loader, test_loader]):
metrics, _, y_true, y_pred = run_stat_epoch(args, model, data_loader, return_pred=True)
all_metrics[dsName] = (metrics, y_true, y_pred)
print('-'*50)
all_metric_names = set()
for dsName in ['train', 'val', 'test']:
metrics, _, _ = all_metrics[dsName]
for k in metrics.keys():
all_metric_names.add(k)
#
print('%25s' % '', end='')
for dsName in ['train', 'val', 'test']:
print('%12s' % dsName, end='')
print()
for mName in all_metric_names:
print('%25s' % mName, end='')
for dsName in ['train', 'val', 'test']:
metrics, _, _ = all_metrics[dsName]
print('%12.5f' % metrics[mName], end='')
print()
#
print('-'*50)
_baseFname = pdb2graph_translator.get_best_model_filename().replace('.pth', '')
for dsName in ['train', 'val', 'test']:
_, y_true, y_pred = all_metrics[dsName]
y_true = np.array(y_true)[:, 0]
y_pred = np.array(y_pred)
np.savetxt(_baseFname+'_yy_%s.txt' % dsName, np.vstack((y_true, y_pred)).T,
header='y_true, y_pred (%s)' % dsName )