DiffPharma/interaction_construct.py at main · sekijima-lab/DiffPharma · GitHub

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
import oddt
import numpy as np
from oddt.interactions import hbonds, hbond_acceptor_donor,  hydrophobic_contacts
from tqdm import tqdm
import argparse
from pathlib import Path

def hbond_create(protein, ligand):
    li_do = hbond_acceptor_donor(protein,ligand)
    li_ac = hbond_acceptor_donor(ligand, protein)

    # ligand pocket interaction id
    po_id_d = [li_do[0][i]['id'] for i in range(len(li_do[0])) if li_do[2][i]==True]
    po_id_a = [li_ac[1][i]['id'] for i in range(len(li_ac[0])) if li_ac[2][i]==True]

    li_id_d = [li_do[1][i]['id'] for i in range(len(li_do[0])) if li_do[2][i]==True]
    li_id_a = [li_ac[0][i]['id'] for i in range(len(li_ac[0])) if li_ac[2][i]==True]

    int_id = []
    if len(li_id_d)>0:
        [int_id.append([li_id_d[i],0,po_id_d[i],0])for i in range(len(li_id_d))]
    if len(li_id_a)>0:
        [int_id.append([0,li_id_a[i],0,po_id_a[i]])for i in range(len(li_id_a))]

    # interaction particle coords
    coords_lido_1 = [(3*li_do[0][i]['coords']+li_do[1][i]['coords'])/4 for i in range(len(li_do[0])) if li_do[2][i]==True]
    coords_lido_2 = [(li_do[0][i]['coords']+li_do[1][i]['coords'])/2 for i in range(len(li_do[0])) if li_do[2][i]==True]
    coords_lido_3 = [(li_do[0][i]['coords']+3*li_do[1][i]['coords'])/4 for i in range(len(li_do[0])) if li_do[2][i]==True]
    coords_liac_1 = [(3*li_ac[0][i]['coords']+li_ac[1][i]['coords'])/4 for i in range(len(li_ac[0])) if li_ac[2][i]==True]
    coords_liac_2 = [(li_ac[0][i]['coords']+li_ac[1][i]['coords'])/2 for i in range(len(li_ac[0])) if li_ac[2][i]==True]
    coords_liac_3 = [(li_ac[0][i]['coords']+3*li_ac[1][i]['coords'])/4 for i in range(len(li_ac[0])) if li_ac[2][i]==True]

    inter_coords=[]
    [inter_coords.append(item) for pair in zip(coords_lido_1, coords_lido_2, coords_lido_3) for item in pair]
    [inter_coords.append(item) for pair in zip(coords_liac_1, coords_liac_2, coords_liac_3) for item in pair]
    inter_coords=np.array(inter_coords)

    #len_do = len(coords_lido_1)*3
    #len_ac = len(coords_liac_1)*3
    #do_ac_id = np.concatenate([np.zeros(len_do,dtype=int),np.ones(len_ac,dtype=int)])
    do_ids = np.tile([0, 1, 1], len(coords_lido_1))
    ac_ids = np.tile([0, 2, 2], len(coords_liac_1))
    do_ac_id = np.concatenate([do_ids, ac_ids])
    inter_one_hot = np.identity(3)[do_ac_id]

    if len(int_id)==0:
        int_id=[]
        inter_coords = np.array([], dtype=np.float32).reshape(0, 3)
        inter_one_hot = np.array([], dtype=np.float32).reshape(0, 3)

    return int_id, inter_coords, inter_one_hot

def hydrophobic_data(protein, ligand):
    hydo = hydrophobic_contacts(protein, ligand)

    # ligand pocket interaction id
    int_id = np.stack((hydo[1]['id'], hydo[0]['id']), axis=1).astype(np.int64)

    # interaction particle coords
    C1 = (2*hydo[0]['coords'] + hydo[1]['coords'])/3
    C2 = (hydo[0]['coords'] + 2*hydo[1]['coords'])/3
    coords = np.array([item for pair in zip(C1,C2) for item in pair])

    # ligand atom type
    mapping = {'SP':0, 'C.ar': 1,'C.3': 2,'C.2': 3,'C.1': 4} # others:5
    atom_type = [x['atomtype'] for x in hydo[1]]
    atom_sp = ['SP']*len(atom_type)
    atom_label = [x for i in range(len(atom_type)) for x in (atom_sp[i], atom_type[i])]
    atom_int = [mapping.get(x, 5) for x in atom_label]
    one_hot = np.identity(6)[atom_int]

    if len(hydo[0])==0:
        int_id=np.array([], dtype=np.float32).reshape(0, 2)
        coords = np.array([], dtype=np.float32).reshape(0, 3)
        one_hot = np.array([], dtype=np.float32).reshape(0, 6)

    return int_id, coords, one_hot

def data_create(file, base_path):
    inter_id_h = []
    inter_mask_h = []
    inter_coords_h = []
    inter_one_hot_h = []
    inter_id_hp = []
    inter_mask_hp = []
    inter_coords_hp = []
    inter_one_hot_hp = []

    pdb_name = [x.split('.pdb_')[0]+'.pdb' for x in file['names']]
    lig_name = [x.split('.pdb_')[1] for x in file['names']]

    for i in tqdm(range(len(file['names']))):
        protein = next(oddt.toolkit.readfile('pdb',str(Path(base_path, pdb_name[i]))))
        ligand = next(oddt.toolkit.readfile('sdf',str(Path(base_path, lig_name[i]))))
        protein.protein = True
        ligand.removeh()

        h_id, h_coords, h_one_hot = hbond_create(protein, ligand)
        hp_id, hp_coords, hp_one_hot = hydrophobic_data(protein, ligand)
        [inter_id_h.append(x) for x in h_id]
        [inter_coords_h.append(x) for x in h_coords]
        [inter_one_hot_h.append(x)for x in h_one_hot]
        [inter_mask_h.append(x) for x in np.full(len(h_coords),i)]
        [inter_id_hp.append(x) for x in hp_id]
        [inter_coords_hp.append(x) for x in hp_coords]
        [inter_one_hot_hp.append(x)for x in hp_one_hot]
        [inter_mask_hp.append(x) for x in np.full(len(hp_coords),i)]

    inter_id_h = np.array(inter_id_h)
    inter_coords_h = np.array(inter_coords_h)
    inter_one_hot_h = np.array(inter_one_hot_h)
    inter_mask_h = np.array(inter_mask_h)
    inter_id_hp = np.array(inter_id_hp)
    inter_coords_hp = np.array(inter_coords_hp)
    inter_one_hot_hp = np.array(inter_one_hot_hp)
    inter_mask_hp = np.array(inter_mask_hp)

    return inter_id_h, inter_coords_h, inter_one_hot_h, inter_mask_h, inter_id_hp, inter_coords_hp, inter_one_hot_hp, inter_mask_hp

def data_save(data_dir, new_dir, pdb_dir, data_type):
    print('=== data creation for ' + data_type + ' ===')

    # read original data without hbond
    file_path = Path(data_dir,data_type + '.npz')
    data_file = np.load(file_path)

    inter_id_h, inter_coords_h, inter_one_hot_h, inter_mask_h, inter_id_hp, inter_coords_hp, inter_one_hot_hp, inter_mask_hp = data_create(data_file, pdb_dir)

    # data save
    out_name = Path(new_dir, data_type +'.npz')

    np.savez(out_name,
            names=data_file['names'],
            lig_coords=data_file['lig_coords'],
            lig_one_hot=data_file['lig_one_hot'],
            lig_mask=data_file['lig_mask'],
            pocket_coords=data_file['pocket_coords'],
            pocket_one_hot=data_file['pocket_one_hot'],
            pocket_mask=data_file['pocket_mask'],
            interh_id=inter_id_h,
            interh_coords = inter_coords_h,
            interh_one_hot = inter_one_hot_h,
            interh_mask=inter_mask_h,
            interhp_id=inter_id_hp,
            interhp_coords = inter_coords_hp,
            interhp_one_hot = inter_one_hot_hp,
            interhp_mask=inter_mask_hp,
            )

def main():
    parser = argparse.ArgumentParser()
    parser.add_argument('--data_dir', type=Path,default=None)
    parser.add_argument('--out_dir', type=Path,default=None)
    parser.add_argument('--pdb_dir', type=Path,default=None)
    parser.add_argument('--skip_existing', action='store_true', default=True)
    args = parser.parse_args()

    args.out_dir.mkdir(exist_ok=args.skip_existing)

    data_type = ['val', 'test', 'train']
    [data_save(args.data_dir, args.out_dir ,args.pdb_dir ,n) for n in data_type]

if __name__ == "__main__":
    main()