massformer/preproc_scripts/prepare_data.py at master · Roestlab/massformer · 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
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
192
193
import numpy as np
import os
import pandas as pd
import ast
import json
from pprint import pprint, pformat
import joblib
from tqdm import tqdm
import argparse

import massformer.data_utils as data_utils
from massformer.data_utils import par_apply_series, par_apply_df_rows, seq_apply_series, seq_apply_df_rows, check_mol_props
from massformer.misc_utils import list_str2float, booltype, tqdm_joblib


def load_df(df_dp, dset_names, num_entries):

    dfs = []
    for dset_name in dset_names:
        dset_fp = os.path.join(df_dp, f"{dset_name}_df.json")
        dset_df = pd.read_json(open(dset_fp, "r", encoding="utf8"))
        dset_df.loc[:, "dset"] = dset_name
        dfs.append(dset_df)
    if num_entries > 0:
        dfs = [
            df.sample(
                n=num_entries,
                replace=False,
                random_state=420) for df in dfs]
    if len(dfs) > 1:
        all_df = pd.concat(dfs, ignore_index=True)
    else:
        all_df = dfs[0]
    all_df = all_df.reset_index(drop=True)

    return all_df

# preprocesses spectra and molecules
# remove compounds/spectra with invalid smiles
# remove compounds/spectra with no bonds or invalid atoms
# this filtering is done here because it would be expensive to do at the
# beginning of training


def preprocess_spec(spec_df):

    # convert smiles to mol and back (for standardization/stereochemistry)
    spec_df.loc[:, "mol"] = par_apply_series(
        spec_df["smiles"], data_utils.mol_from_smiles)
    spec_df.loc[:, "smiles"] = par_apply_series(
        spec_df["mol"], data_utils.mol_to_smiles)
    spec_df = spec_df.dropna(subset=["mol", "smiles"])
    # check atom types, number of bonds, neutral charge
    spec_df = check_mol_props(spec_df)
    # enumerate smiles to create molecule ids
    smiles_set = set(spec_df["smiles"])
    print("> num_smiles", len(smiles_set))
    smiles_to_mid = {smiles: i for i, smiles in enumerate(sorted(smiles_set))}
    spec_df.loc[:, "mol_id"] = spec_df["smiles"].replace(smiles_to_mid)

    # extract peak info (still represented as str)
    spec_df.loc[:, "peaks"] = par_apply_series(
        spec_df["peaks"], data_utils.parse_peaks_str)
    # get mz resolution
    spec_df.loc[:, "res"] = par_apply_series(
        spec_df["peaks"], data_utils.get_res)
    # standardize the instrument type and frag_mode
    inst_type, frag_mode = seq_apply_df_rows(
        spec_df, data_utils.parse_inst_info)
    spec_df.loc[:, "inst_type"] = inst_type
    spec_df.loc[:, "frag_mode"] = frag_mode
    # standardize ce
    spec_df.loc[:, "ace"] = par_apply_series(
        spec_df["col_energy"], data_utils.parse_ace_str)
    spec_df.loc[:, "nce"] = par_apply_series(
        spec_df["col_energy"], data_utils.parse_nce_str)
    spec_df = spec_df.drop(columns=["col_energy"])
    # standardise prec_type
    spec_df.loc[:, "prec_type"] = par_apply_series(
        spec_df["prec_type"], data_utils.parse_prec_type_str)
    # convert prec_mz
    spec_df.loc[:, "prec_mz"] = pd.to_numeric(
        spec_df["prec_mz"], errors="coerce")
    spec_df = spec_df.astype({"prec_mz": float})
    # convert ion_mode
    spec_df.loc[:, "ion_mode"] = par_apply_series(
        spec_df["ion_mode"], data_utils.parse_ion_mode_str)
    # convert peaks to float
    spec_df.loc[:, "peaks"] = par_apply_series(
        spec_df["peaks"], data_utils.convert_peaks_to_float)
    # get retention index
    spec_df.loc[:, "ri"] = par_apply_series(
        spec_df["ri"], data_utils.parse_ri_str)

    # remove columns from spec_df
    spec_df = spec_df[["spec_id",
                       "mol_id",
                       "prec_type",
                       "inst_type",
                       "frag_mode",
                       "spec_type",
                       "ion_mode",
                       "dset",
                       "col_gas",
                       "res",
                       "ace",
                       "nce",
                       "prec_mz",
                       "peaks",
                       "ri"]]
    # relabel spec_id (this is to make it unique across datasets)
    spec_df.loc[:, "spec_id"] = np.arange(spec_df.shape[0])

    # get mol df
    mol_df = pd.DataFrame(zip(sorted(smiles_set), list(
        range(len(smiles_set)))), columns=["smiles", "mol_id"])
    mol_df.loc[:, "mol"] = par_apply_series(
        mol_df["smiles"], data_utils.mol_from_smiles)
    mol_df.loc[:, "inchikey_s"] = par_apply_series(
        mol_df["mol"], data_utils.mol_to_inchikey_s)
    mol_df.loc[:, "scaffold"] = par_apply_series(
        mol_df["mol"], data_utils.get_murcko_scaffold)
    mol_df.loc[:, "formula"] = par_apply_series(
        mol_df["mol"], data_utils.mol_to_formula)
    mol_df.loc[:, "inchi"] = par_apply_series(
        mol_df["mol"], data_utils.mol_to_inchi)
    mol_df.loc[:, "mw"] = par_apply_series(
        mol_df["mol"], lambda mol: data_utils.mol_to_mol_weight(mol, exact=False))
    mol_df.loc[:, "exact_mw"] = par_apply_series(
        mol_df["mol"], lambda mol: data_utils.mol_to_mol_weight(mol, exact=True))

    # remove invalid mols and corresponding spectra
    all_mol_id = set(mol_df["mol_id"])
    mol_df = mol_df.dropna(subset=["mol"], axis=0)
    bad_mol_id = all_mol_id - set(mol_df["mol_id"])
    print("> bad_mol_id", len(bad_mol_id))
    spec_df = spec_df[~spec_df["mol_id"].isin(bad_mol_id)]

    # reset indices
    spec_df = spec_df.reset_index(drop=True)
    mol_df = mol_df.reset_index(drop=True)

    return spec_df, mol_df


if __name__ == "__main__":

    parser = argparse.ArgumentParser()
    parser.add_argument("--df_dp", type=str, default="data/df")
    parser.add_argument("--dset_names", type=str, default="nist,mb_na")
    parser.add_argument("--proc_dp", type=str, default="data/proc")
    parser.add_argument("--num_entries", type=int, default=-1)
    flags = parser.parse_args()

    os.makedirs(flags.proc_dp, exist_ok=True)
    data_dp = flags.proc_dp

    spec_df_fp = os.path.join(data_dp, "spec_df.pkl")
    mol_df_fp = os.path.join(data_dp, "mol_df.pkl")

    print("> creating new spec_df, mol_df")
    assert os.path.isdir(flags.df_dp), flags.df_dp
    dset_names = flags.dset_names.split(",")
    all_df = load_df(flags.df_dp, dset_names, flags.num_entries)

    spec_df, mol_df = preprocess_spec(all_df)

    # save everything to file
    spec_df.to_pickle(spec_df_fp)
    mol_df.to_pickle(mol_df_fp)

    print(spec_df.shape)
    print(spec_df.isna().sum())
    print(mol_df.shape)
    print(mol_df.isna().sum())

    # export smiles (.txt, cfm) and inchi (.tsv, classyfire)
    smiles_df = mol_df[["mol_id", "smiles"]]
    smiles_df.to_csv(
        os.path.join(
            data_dp,
            "all_smiles.txt"),
        sep=" ",
        header=False,
        index=False)
    inchi_df = mol_df[["mol_id", "inchi"]]
    inchi_df.to_csv(
        os.path.join(
            data_dp,
            "all_inchi.tsv"),
        sep="\t",
        header=False,
        index=False)