|
1
|
|
|
#! /usr/bin/env python |
|
|
|
|
|
|
2
|
|
|
# |
|
3
|
|
|
# Copyright (C) 2016 Rich Lewis <[email protected]> |
|
4
|
|
|
# License: 3-clause BSD |
|
5
|
|
|
|
|
6
|
|
|
""" |
|
7
|
|
|
## skchem.standardizers.chemaxon |
|
8
|
|
|
|
|
9
|
|
|
Module wrapping ChemAxon Standardizer. Must have standardizer installed and |
|
10
|
|
|
license activated. |
|
11
|
|
|
""" |
|
12
|
|
|
|
|
13
|
|
|
import os |
|
14
|
|
|
import re |
|
15
|
|
|
from tempfile import NamedTemporaryFile |
|
16
|
|
|
import subprocess |
|
17
|
|
|
import logging |
|
18
|
|
|
|
|
19
|
|
|
logger = logging.getLogger(__name__) |
|
|
|
|
|
|
20
|
|
|
|
|
21
|
|
|
import numpy as np |
|
|
|
|
|
|
22
|
|
|
import pandas as pd |
|
|
|
|
|
|
23
|
|
|
|
|
24
|
|
|
from .. import core |
|
25
|
|
|
from .. import io |
|
26
|
|
|
|
|
27
|
|
|
# ideally we will programatically build this file, but for now just use it. |
|
28
|
|
|
DEFAULT_CONFIG = os.path.join(os.path.dirname(__file__), 'default_config.xml') |
|
29
|
|
|
|
|
30
|
|
|
class ChemAxonStandardizer(object): |
|
31
|
|
|
|
|
32
|
|
|
""" Object wrapping the ChemAxon Standardizer, for standardizing molecules. |
|
33
|
|
|
|
|
34
|
|
|
Args: |
|
35
|
|
|
config_path (str): |
|
36
|
|
|
The path of the config_file. If None, use the default one. |
|
37
|
|
|
""" |
|
38
|
|
|
def __init__(self, config_path=None, warn_on_fail=True, error_on_fail=False, |
|
39
|
|
|
keep_failed=False): |
|
|
|
|
|
|
40
|
|
|
|
|
41
|
|
|
if not config_path: |
|
42
|
|
|
config_path = DEFAULT_CONFIG |
|
43
|
|
|
self.config_path = config_path |
|
44
|
|
|
self.keep_failed = keep_failed |
|
45
|
|
|
self.error_on_fail = error_on_fail |
|
46
|
|
|
self.warn_on_fail = warn_on_fail |
|
47
|
|
|
|
|
48
|
|
|
def transform(self, obj): |
|
49
|
|
|
|
|
50
|
|
|
""" Standardize compounds. |
|
51
|
|
|
|
|
52
|
|
|
Args: |
|
53
|
|
|
obj (str, skchem.Mol, pd.Series or pd.DataFrame): |
|
54
|
|
|
The object to standardize as either smiles as a string, Mol, or |
|
55
|
|
|
a series or dataframe of these. The object to standardize. |
|
56
|
|
|
|
|
57
|
|
|
Returns: |
|
58
|
|
|
skchem.Mol or pd.Series or pd.DataFrame: |
|
59
|
|
|
The standardized molecule, or molecules as a series or |
|
60
|
|
|
dataframe. |
|
61
|
|
|
""" |
|
62
|
|
|
|
|
63
|
|
|
if isinstance(obj, core.Mol): |
|
64
|
|
|
return self._transform_mol(obj) |
|
65
|
|
|
elif isinstance(obj, pd.Series): |
|
66
|
|
|
return self._transform_ser(obj) |
|
67
|
|
|
elif isinstance(obj, pd.DataFrame): |
|
68
|
|
|
res = self._transform_ser(obj.structure) |
|
69
|
|
|
return res.to_frame(name='structure').join(obj.drop('structure', axis=1)) |
|
70
|
|
|
|
|
71
|
|
|
else: |
|
72
|
|
|
raise NotImplementedError |
|
73
|
|
|
|
|
74
|
|
|
def _transform_mol(self, mol): |
|
75
|
|
|
mol = pd.DataFrame([mol], index=[mol.name], columns=['structure']) |
|
76
|
|
|
return self.transform(mol).structure.iloc[0] |
|
77
|
|
|
|
|
78
|
|
|
def _transform_mols(self, X, by='sdf'): |
|
|
|
|
|
|
79
|
|
|
|
|
80
|
|
|
with NamedTemporaryFile() as f_in, NamedTemporaryFile() as f_out: |
|
81
|
|
|
getattr(io, 'write_' + by)(X, f_in.name) |
|
82
|
|
|
errs = self._transform_file(f_in.name, f_out.name) |
|
83
|
|
|
out = io.read_sdf(f_out.name).structure |
|
84
|
|
|
return out, errs |
|
85
|
|
|
|
|
86
|
|
|
def _transform_smis(self, X): |
|
|
|
|
|
|
87
|
|
|
with NamedTemporaryFile() as f_in, NamedTemporaryFile() as f_out: |
|
88
|
|
|
X.to_csv(f_in.name, header=None, index=None) |
|
89
|
|
|
errs = self._transform_file(f_in.name, f_out.name) |
|
90
|
|
|
out = io.read_sdf(f_out.name).structure |
|
91
|
|
|
return out, errs |
|
92
|
|
|
|
|
93
|
|
|
def _transform_file(self, f_in, f_out): |
|
94
|
|
|
args = ['standardize', f_in, |
|
95
|
|
|
'-c', self.config_path, |
|
|
|
|
|
|
96
|
|
|
'-f', 'sdf', |
|
|
|
|
|
|
97
|
|
|
'-o', f_out, |
|
|
|
|
|
|
98
|
|
|
'--ignore-error'] |
|
|
|
|
|
|
99
|
|
|
logger.debug('Running %s', ' '.join(args)) |
|
100
|
|
|
sub = subprocess.Popen(args, stderr=subprocess.PIPE) |
|
101
|
|
|
errs = sub.stderr.read().decode('ascii') |
|
102
|
|
|
if len(errs): |
|
103
|
|
|
logger.debug('stderr from Standardizer: \n%s', errs) |
|
104
|
|
|
errs = errs.strip().split('\n') |
|
105
|
|
|
errs = [re.findall('No. ([0-9]+):', err) for err in errs] |
|
106
|
|
|
errs = [int(err[0]) - 1 for err in errs if len(err)] |
|
107
|
|
|
return errs |
|
108
|
|
|
|
|
109
|
|
|
def _transform_ser(self, X, y=None): |
|
|
|
|
|
|
110
|
|
|
|
|
111
|
|
|
# TODO: try using different serializations |
|
|
|
|
|
|
112
|
|
|
if isinstance(X.iloc[0], core.Mol): |
|
113
|
|
|
out, errs = self._transform_mols(X) |
|
114
|
|
|
elif isinstance(X.iloc[0], str): |
|
115
|
|
|
out, errs = self._transform_smis(X) |
|
116
|
|
|
if errs: |
|
117
|
|
|
out.index = X.index.delete(errs) |
|
118
|
|
|
for err in errs: |
|
119
|
|
|
err = X.index[err] |
|
120
|
|
|
if self.error_on_fail: |
|
121
|
|
|
raise ValueError('{} failed to standardize'.format(err)) |
|
122
|
|
|
if self.warn_on_fail: |
|
123
|
|
|
logger.warn('%s failed to standardize', err) |
|
124
|
|
|
else: |
|
125
|
|
|
out.index = X.index |
|
126
|
|
|
if self.keep_failed: |
|
127
|
|
|
out_c = X.copy() |
|
128
|
|
|
out_c.loc[out.index] = out |
|
129
|
|
|
return out_c |
|
130
|
|
|
else: |
|
131
|
|
|
return out |
|
132
|
|
|
|
richlewis42 /
scikit-chem
Loading history...
Cyclic imports may cause partly loaded modules to be returned. This might lead to unexpected runtime behavior which is hard to debug.