-
Notifications
You must be signed in to change notification settings - Fork 6
/
Synthesis.py
178 lines (160 loc) · 9.33 KB
/
Synthesis.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
from itertools import permutations
import pandas as pd
import json
from rdkit import Chem
import torch
from torch import nn
import sklearn
import dgl
from dgllife.utils import smiles_to_bigraph, WeaveAtomFeaturizer, CanonicalBondFeaturizer
from functools import partial
from scripts.dataset import combine_reactants, get_bonds, get_adm
from scripts.utils import init_featurizer, load_model, pad_atom_distance_matrix, predict
from scripts.get_edit import get_bg_partition, combined_edit
from LocalTemplate.template_collector import Collector
atom_types = ['C', 'N', 'O', 'S', 'F', 'Si', 'P', 'Cl', 'Br', 'Mg', 'Na', 'Ca', 'Fe',
'As', 'Al', 'I', 'B', 'V', 'K', 'Tl', 'Yb', 'Sb', 'Sn', 'Ag', 'Pd', 'Co', 'Se', 'Ti',
'Zn', 'H', 'Li', 'Ge', 'Cu', 'Au', 'Ni', 'Cd', 'In', 'Mn', 'Zr', 'Cr', 'Pt', 'Hg', 'Pb',
'W', 'Ru', 'Nb', 'Re', 'Te', 'Rh', 'Ta', 'Tc', 'Ba', 'Bi', 'Hf', 'Mo', 'U', 'Sm', 'Os', 'Ir',
'Ce', 'Gd', 'Ga', 'Cs']
def demap(smiles):
mol = Chem.MolFromSmiles(smiles)
[atom.SetAtomMapNum(0) for atom in mol.GetAtoms()]
return Chem.MolToSmiles(mol)
class localtransform():
def __init__(self, dataset, device='cuda:0'):
self.data_dir = 'data/%s' % dataset
self.config_path = 'data/configs/default_config'
self.model_path = 'models/LocalTransform_%s.pth' % dataset
self.device = torch.device(device) if torch.cuda.is_available() else torch.device('cpu')
self.args = {'data_dir': self.data_dir, 'model_path': self.model_path, 'config_path': self.config_path, 'device': self.device, 'mode': 'test'}
self.template_dicts, self.template_infos = self.load_templates()
self.model, self.graph_function = self.init_model()
def load_templates(self):
template_dicts = {}
for site in ['real', 'virtual']:
template_df = pd.read_csv('%s/%s_templates.csv' % (self.data_dir, site))
template_dict = {template_df['Class'][i]: template_df['Template'][i].split('_') for i in template_df.index}
print ('loaded %s %s templates' % (len(template_dict), site))
template_dicts[site[0]] = template_dict
template_infos = pd.read_csv('%s/template_infos.csv' % self.data_dir)
template_infos = {template_infos['Template'][i]: {
'edit_site': eval(template_infos['edit_site'][i]),
'change_H': eval(template_infos['change_H'][i]),
'change_C': eval(template_infos['change_C'][i]),
'change_S': eval(template_infos['change_S'][i])} for i in template_infos.index}
return template_dicts, template_infos
def init_model(self):
self.args = init_featurizer(self.args)
model = load_model(self.args)
model.eval()
smiles_to_graph = partial(smiles_to_bigraph, add_self_loop=True)
node_featurizer = WeaveAtomFeaturizer(atom_types=atom_types)
edge_featurizer = CanonicalBondFeaturizer(self_loop=True)
graph_function = lambda s: smiles_to_graph(s, node_featurizer = node_featurizer, edge_featurizer = edge_featurizer, canonical_atom_order = False)
return model, graph_function
def make_inference(self, reactant_list, topk=5):
fgraphs = []
dgraphs = []
for smiles in reactant_list:
mol = Chem.MolFromSmiles(smiles)
fgraph = self.graph_function(smiles)
dgraph = {'atom_distance_matrix': get_adm(mol), 'bonds':get_bonds(smiles)}
dgraph['v_bonds'], dgraph['r_bonds'] = dgraph['bonds']
fgraphs.append(fgraph)
dgraphs.append(dgraph)
bg = dgl.batch(fgraphs)
bg.set_n_initializer(dgl.init.zero_initializer)
bg.set_e_initializer(dgl.init.zero_initializer)
adm_lists = [graph['atom_distance_matrix'] for graph in dgraphs]
adms = pad_atom_distance_matrix(adm_lists)
bonds_dicts = {'virtual': [torch.from_numpy(graph['v_bonds']).long() for graph in dgraphs], 'real': [torch.from_numpy(graph['r_bonds']).long() for graph in dgraphs]}
with torch.no_grad():
pred_VT, pred_RT, _, _, pred_VI, pred_RI, attentions = predict(self.args, self.model, bg, adms, bonds_dicts)
pred_VT = nn.Softmax(dim=1)(pred_VT)
pred_RT = nn.Softmax(dim=1)(pred_RT)
v_sep, r_sep = get_bg_partition(bg, bonds_dicts)
start_v, start_r = 0, 0
predictions = []
for i, (reactant) in enumerate(reactant_list):
end_v, end_r = v_sep[i], r_sep[i]
virtual_bonds, real_bonds = bonds_dicts['virtual'][i].numpy(), bonds_dicts['real'][i].numpy()
pred_vi, pred_ri = pred_VI[i].cpu(), pred_RI[i].cpu()
pred_v, pred_r = pred_VT[start_v:end_v], pred_RT[start_r:end_r]
prediction = combined_edit(virtual_bonds, real_bonds, pred_vi, pred_ri, pred_v, pred_r, topk*10)
predictions.append(prediction)
start_v = end_v
start_r = end_r
return predictions
def predict_product(self, reactant_list, topk=5, verbose=0):
if isinstance(reactant_list, str):
reactant_list = [reactant_list]
predictions = self.make_inference(reactant_list, topk)
results_df = {'Reactants' : []}
results_dict = {}
for k in range(topk):
results_df['Top-%d' % (k+1)] = []
for reactant, prediction in zip(reactant_list, predictions):
pred_types, pred_sites, scores = prediction
collector = Collector(reactant, self.template_infos, 'nan', False, verbose = verbose > 1)
for k, (pred_type, pred_site, score) in enumerate(zip(pred_types, pred_sites, scores)):
template, H_code, C_code, S_code, action = self.template_dicts[pred_type][pred_site[1]]
pred_site = pred_site[0]
if verbose > 0:
print ('%dth prediction:' % (k+1), template, action, pred_site, score)
collector.collect(template, H_code, C_code, S_code, action, pred_site, score)
if len(collector.predictions) >= topk:
break
sorted_predictions = [k for k, v in sorted(collector.predictions.items(), key=lambda item: -item[1]['score'])]
results_df['Reactants'].append(Chem.MolFromSmiles(reactant))
results_dict[reactant] = {}
for k in range(topk):
p = sorted_predictions[k] if len(sorted_predictions)>k else ''
results_dict[reactant]['Top-%d' % (k+1)] = collector.predictions[p]
results_dict[reactant]['Top-%d' % (k+1)]['product'] = p
results_df['Top-%d' % (k+1)].append(Chem.MolFromSmiles(p))
results_df = pd.DataFrame(results_df)
return results_df, results_dict
# def predict_products(self, args, reactant_list, model, graph_functions, template_dicts, template_infos, product = None, reagents = 'nan', top_k = 5, collect_n = 100, verbose = 0, sep = False):
# model.eval()
# if reagents != 'nan':
# smiles = reactant + '.' + reagents
# else:
# smiles = reactant
# dglgraph = graph_functions(smiles)
# adms = pad_atom_distance_matrix([get_adm(Chem.MolFromSmiles(smiles))])
# v_bonds, r_bonds = get_bonds(smiles)
# bonds_dicts = {'virtual': [torch.from_numpy(v_bonds).long()], 'real': [torch.from_numpy(r_bonds).long()]}
# with torch.no_grad():
# pred_VT, pred_RT, _, _, pred_VI, pred_RI, attentions = predict(args, model, dglgraph, adms, bonds_dicts)
# pred_v = nn.Softmax(dim=1)(pred_VT)
# pred_r = nn.Softmax(dim=1)(pred_RT)
# pred_vi = pred_VI[0].cpu()
# pred_ri = pred_RI[0].cpu()
# pred_types, pred_sites, pred_scores = combined_edit(v_bonds, r_bonds, pred_vi, pred_ri, pred_v, pred_r, collect_n)
# collector = Collector(reactant, template_infos, reagents, sep, verbose = verbose > 1)
# for k, (pred_type, pred_site, score) in enumerate(zip(pred_types, pred_sites, pred_scores)):
# template, H_code, C_code, S_code, action = template_dicts[pred_type][pred_site[1]]
# pred_site = pred_site[0]
# if verbose > 0:
# print ('%sth prediction:' % k, template, action, pred_site, score)
# collector.collect(template, H_code, C_code, S_code, action, pred_site, score)
# if len(collector.predictions) >= top_k:
# break
# sort_predictions = [k for k, v in sorted(collector.predictions.items(), key=lambda item: -item[1]['score'])]
# reactant = demap(reactant)
# if product != None:
# correct_at = False
# product = demap(product)
# results_dict = {'Reactants' : demap(reactant)}
# results_df = pd.DataFrame({'Reactants' : [Chem.MolFromSmiles(reactant)]})
# for k, p in enumerate(sort_predictions):
# results_dict['Top-%d' % (k+1)] = collector.predictions[p]
# results_dict['Top-%d' % (k+1)]['product'] = p
# results_df['Top-%d' % (k+1)] = [Chem.MolFromSmiles(p)]
# if product != None:
# if set(p.split('.')).intersection(set(product.split('.'))):
# correct_at = k+1
# if product != None:
# results_df['Correct at'] = correct_at
# return results_df, results_dict