-
Notifications
You must be signed in to change notification settings - Fork 2
/
prepare_data.py
191 lines (168 loc) · 8.11 KB
/
prepare_data.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
# prepare_data.py
#
# This is the code for preparing training data.
#
# Author: Shugang Zhang
# Created: Wednesday, Aug 4th, 2021
# Last update: Sunday, Dec 26th, 2021
import pandas as pd
import numpy as np
import os
import json, pickle
from collections import OrderedDict
from rdkit import Chem
from rdkit.Chem import MolFromSmiles
import networkx as nx
from utils import *
def atom_features(atom):
return np.array(one_of_k_encoding_unk(atom.GetSymbol(),['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', 'Unknown']) +
one_of_k_encoding(atom.GetDegree(), [0, 1, 2, 3, 4, 5, 6,7,8,9,10]) +
one_of_k_encoding_unk(atom.GetTotalNumHs(), [0, 1, 2, 3, 4, 5, 6,7,8,9,10]) +
one_of_k_encoding_unk(atom.GetImplicitValence(), [0, 1, 2, 3, 4, 5, 6,7,8,9,10]) +
[atom.GetIsAromatic()])
def one_of_k_encoding(x, allowable_set):
if x not in allowable_set:
raise Exception("input {0} not in allowable set{1}:".format(x, allowable_set))
return list(map(lambda s: x == s, allowable_set))
def one_of_k_encoding_unk(x, allowable_set):
"""Maps inputs not in the allowable set to the last element."""
if x not in allowable_set:
x = allowable_set[-1]
return list(map(lambda s: x == s, allowable_set))
def smile_to_graph(smile):
mol = Chem.MolFromSmiles(smile)
if (mol is None):
print("bad smile:", smile)
else:
# 1.num of atoms
num_atoms = mol.GetNumAtoms()
# 2.features
features = []
for atom in mol.GetAtoms():
feature = atom_features(atom)
features.append( feature / sum(feature) )
# 3.edges
edges = []
for bond in mol.GetBonds():
edges.append([bond.GetBeginAtomIdx(), bond.GetEndAtomIdx()])
g = nx.Graph(edges).to_directed()
edge_index = []
for e1, e2 in g.edges:
edge_index.append([e1, e2])
return num_atoms, features, edge_index
def seq_cat(prot):
x = np.zeros(max_seq_len)
for i, ch in enumerate(prot[:max_seq_len]):
x[i] = seq_dict[ch]
return x
# from DeepDTA data
all_prots = []
datasets = ['kiba', 'davis']
for dataset in datasets:
print('convert data from DeepDTA for ', dataset)
fpath = 'data/' + dataset + '/'
train_fold = json.load(open(fpath + "folds/train_fold_setting1.txt")) # len=5
train_fold = [ee for e in train_fold for ee in e] # davis len=25046
valid_fold = json.load(open(fpath + "folds/test_fold_setting1.txt")) # davis len=5010
ligands = json.load(open(fpath + "ligands_can.txt"), object_pairs_hook=OrderedDict) # davis len=68
proteins = json.load(open(fpath + "proteins.txt"), object_pairs_hook=OrderedDict) # davis len=442
affinity = pickle.load(open(fpath + "Y", "rb"), encoding='latin1') # davis len=68
drugs = []
prots = []
for d in ligands.keys():
print(d)
lg = Chem.MolToSmiles(Chem.MolFromSmiles(ligands[d]), isomericSmiles=True)
drugs.append(lg) # loading drugs
for t in proteins.keys():
prots.append(proteins[t]) # loading proteins
if dataset == 'davis':
affinity = [-np.log10(y/1e9) for y in affinity]
affinity = np.asarray(affinity) # affinity shape=(68 drug,442 prot)
opts = ['train','test']
for opt in opts:
rows, cols = np.where(np.isnan(affinity)==False) # not NAN
if opt == 'train':
rows, cols = rows[train_fold], cols[train_fold] # train fold包含了drug和prot的index信息
elif opt == 'test':
rows, cols = rows[valid_fold], cols[valid_fold]
with open('data/' + dataset + '_' + opt + '.csv', 'w') as f:
f.write('compound_iso_smiles,target_sequence,affinity\n')
for pair_ind in range(len(rows)):
ls = []
ls += [ drugs[rows[pair_ind]] ]
ls += [ prots[cols[pair_ind]] ]
ls += [ affinity[rows[pair_ind],cols[pair_ind]] ]
f.write(','.join(map(str,ls)) + '\n') # csv format
print('\ndataset:', dataset)
print('train_fold:', len(train_fold))
print('test_fold:', len(valid_fold))
print('len(set(drugs)),len(set(prots)):', len(set(drugs)), len(set(prots)))
all_prots += list(set(prots))
seq_voc = "ABCDEFGHIKLMNOPQRSTUVWXYZ"
seq_dict = {v:(i+1) for i,v in enumerate(seq_voc)} # encode alphabet from 1
seq_dict_len = len(seq_dict)
max_seq_len = 1000
# 2.create graph for all SMILES
print("\nCreating graph for all SMILES...")
compound_iso_smiles = []
for dt_name in ['kiba','davis']:
opts = ['train','test']
for opt in opts:
df = pd.read_csv('data/' + dt_name + '_' + opt + '.csv')
compound_iso_smiles += list(df['compound_iso_smiles'])
compound_iso_smiles = set(compound_iso_smiles)
smile_graph = {}
for smile in compound_iso_smiles:
g = smile_to_graph(smile)
smile_graph[smile] = g
datasets = ['davis','kiba']
# convert to PyTorch data format
for dataset in datasets:
# read in train file
df = pd.read_csv('data/' + dataset + '_train.csv')
drugs, prots, Y = list(df['compound_iso_smiles']), list(df['target_sequence']), list(df['affinity'])
XT = [seq_cat(t) for t in prots]
drugs, prots, Y = np.asarray(drugs), np.asarray(XT), np.asarray(Y)
# shuffle
idx = np.random.permutation(len(drugs))
drugs = drugs[idx]
prots = prots[idx]
Y = Y[idx]
portion = int(0.2 * len(drugs))
# 5-fold, train&valid sets
for fold in range(5):
processed_data_file_train = 'data/processed/' + dataset + '_fold' + str(fold+1) + '_train.pt'
processed_data_file_valid = 'data/processed/' + dataset + '_fold' + str(fold+1) + '_valid.pt'
# splitting
if ((not os.path.isfile(processed_data_file_train)) or (not os.path.isfile(processed_data_file_valid))):
valid_drugs = drugs[fold * portion:(fold + 1) * portion]
train_drugs = np.delete(drugs, range(fold * portion, (fold + 1) * portion))
valid_prots = prots[fold * portion:(fold + 1) * portion]
train_prots = np.delete(prots, range(fold * portion, (fold + 1) * portion), axis=0)
valid_Y = Y[fold * portion:(fold + 1) * portion]
train_Y = np.delete(Y, range(fold * portion, (fold + 1) * portion))
print("shape:", train_drugs.shape, train_prots.shape, train_Y.shape)
# make data PyTorch Geometric ready
print('preparing', dataset + '_train.pt in pytorch format!')
train_data = TestbedDataset(root='data', dataset=dataset + '_fold' + str(fold + 1) + '_train',
xd=train_drugs, xt=train_prots, y=train_Y, smile_graph=smile_graph)
print('preparing', dataset + '_valid.pt in pytorch format!')
valid_data = TestbedDataset(root='data', dataset=dataset + '_fold' + str(fold + 1) + '_valid',
xd=valid_drugs, xt=valid_prots, y=valid_Y, smile_graph=smile_graph)
print(processed_data_file_train, processed_data_file_valid, 'are created successfully.')
else:
print(processed_data_file_train, processed_data_file_valid, 'are already created.')
# test set
processed_data_file_test = 'data/processed/' + dataset + '_test.pt'
if not os.path.isfile(processed_data_file_test):
df = pd.read_csv('data/' + dataset + '_test.csv')
drugs, prots, Y = list(df['compound_iso_smiles']), list(df['target_sequence']), list(df['affinity'])
XT = [seq_cat(t) for t in prots]
drugs, prots, Y = np.asarray(drugs), np.asarray(XT), np.asarray(Y)
# make data PyTorch Geometric ready
print('preparing ', dataset + '_test.pt in pytorch format!')
valid_data = TestbedDataset(root='data', dataset=dataset + '_test',
xd=drugs, xt=prots, y=Y, smile_graph=smile_graph)
print(processed_data_file_test, 'created successfully.')
else:
print(processed_data_file_test, 'are already created.')