1
|
|
|
#! /usr/bin/env python |
2
|
|
|
# |
3
|
|
|
# Copyright (C) 2015-2016 Rich Lewis <[email protected]> |
4
|
|
|
# License: 3-clause BSD |
5
|
|
|
|
6
|
1 |
|
""" |
7
|
|
|
# skchem.filters |
8
|
|
|
|
9
|
|
|
Chemical filters are defined. |
10
|
|
|
|
11
|
|
|
""" |
12
|
|
|
|
13
|
1 |
|
import pandas as pd |
|
|
|
|
14
|
|
|
|
15
|
1 |
|
from ..base import BaseTransformer, Transformer |
16
|
1 |
|
from .. import core |
17
|
1 |
|
from ..utils import iterable_to_series, Defaults, optional_second_method |
18
|
|
|
|
19
|
|
|
|
20
|
1 |
|
def not_all(x): |
|
|
|
|
21
|
|
|
""" Not all x """ |
22
|
|
|
return not all(x) |
23
|
|
|
|
24
|
|
|
|
25
|
1 |
|
def not_any(x): |
|
|
|
|
26
|
|
|
""" Not any x """ |
27
|
1 |
|
return not any(x) |
28
|
|
|
|
29
|
|
|
|
30
|
1 |
|
def identity(x): |
|
|
|
|
31
|
|
|
""" The identity """ |
32
|
|
|
return x |
33
|
|
|
|
34
|
|
|
|
35
|
1 |
|
AGGS = Defaults(defaults={ |
36
|
|
|
'none': identity, |
37
|
|
|
'any': any, |
38
|
|
|
'all': all, |
39
|
|
|
'not all': not_all, |
40
|
|
|
'not any': not_any |
41
|
|
|
}) |
42
|
|
|
|
43
|
|
|
|
44
|
1 |
|
class BaseFilter(BaseTransformer): |
45
|
|
|
|
46
|
|
|
""" The base Filter class. """ |
47
|
|
|
|
48
|
1 |
|
def __init__(self, agg='any', **kwargs): |
49
|
1 |
|
self._agg = None |
50
|
1 |
|
super(BaseFilter, self).__init__(**kwargs) |
51
|
1 |
|
self.agg = agg |
52
|
|
|
|
53
|
1 |
|
def axes_names(self): |
54
|
|
|
return 'batch', self.columns.name |
55
|
|
|
|
56
|
1 |
|
@property |
57
|
|
|
def agg(self): |
58
|
|
|
""" callable: The aggregate function to use. String aliases |
59
|
|
|
for `'any'`, `'not any'`, 'all', `'not all'` are available.""" |
60
|
1 |
|
return self._agg |
61
|
|
|
|
62
|
1 |
|
@agg.setter |
63
|
|
|
def agg(self, val): |
|
|
|
|
64
|
1 |
|
self._agg = AGGS.get(val) |
65
|
|
|
|
66
|
1 |
|
@property |
67
|
|
|
def columns(self): |
68
|
|
|
|
69
|
|
|
""" pd.Index: The column index to use. """ |
70
|
|
|
|
71
|
1 |
|
return pd.Index([self.__class__.__name__]) |
72
|
|
|
|
73
|
1 |
|
def _mask(self, mols=None, res=None, neg=False): |
74
|
|
|
|
75
|
|
|
""" Generate a mask from molecules, or from their result after transform. |
76
|
|
|
|
77
|
|
|
Args: |
78
|
|
|
mols (pd.Series<skchem.Mol>): |
79
|
|
|
The molecules to use to generate the mask. |
80
|
|
|
res (pd.Series): |
81
|
|
|
The result of a transform. Overrides mols. |
82
|
|
|
neg (bool): |
83
|
|
|
Whether the mask should be inversed. |
84
|
|
|
|
85
|
|
|
Returns: |
86
|
|
|
pd.Series<bool> |
87
|
|
|
""" |
88
|
|
|
|
89
|
1 |
|
res = self.transform(mols, agg=False) if res is None else res |
90
|
1 |
|
res = (res != 0) & pd.notnull(res) |
91
|
1 |
|
if isinstance(res, pd.Series) and isinstance(mols, core.Mol): |
92
|
1 |
|
res = self.agg(res) |
|
|
|
|
93
|
1 |
|
if isinstance(res, pd.DataFrame): |
94
|
1 |
|
res = res.apply(self.agg, axis=1) |
95
|
1 |
|
return res == 0 if neg else res |
96
|
|
|
|
97
|
1 |
|
@optional_second_method |
98
|
1 |
|
def transform(self, mols, agg=True, **kwargs): |
99
|
|
|
|
100
|
|
|
# transform takes additional optional kwarg `agg`, that specifies to |
101
|
|
|
# transform to the aggregated value or return the full series. |
102
|
|
|
|
103
|
1 |
|
if agg: |
104
|
1 |
|
return self._mask(mols) |
105
|
|
|
else: |
106
|
1 |
|
return super(BaseFilter, self).transform(mols, **kwargs) |
107
|
|
|
|
108
|
1 |
|
def filter(self, mols, y=None, neg=False): |
|
|
|
|
109
|
|
|
|
110
|
1 |
|
mask = self._mask(mols=mols, neg=neg) |
111
|
|
|
|
112
|
1 |
|
if isinstance(mols, core.Mol): |
113
|
|
|
return mols if mask else None |
114
|
|
|
|
115
|
1 |
|
elif not isinstance(mols, pd.Series): |
116
|
1 |
|
mols = iterable_to_series(mols) |
117
|
|
|
|
118
|
1 |
|
if y is None: |
119
|
1 |
|
return mols[mask] |
120
|
|
|
else: |
121
|
|
|
return mols[mask], y[mask] |
122
|
|
|
|
123
|
|
|
|
124
|
1 |
|
class Filter(BaseFilter, Transformer): |
125
|
|
|
""" Filter base class. |
126
|
|
|
|
127
|
|
|
Examples: |
128
|
|
|
|
129
|
|
|
>>> import skchem |
130
|
|
|
|
131
|
|
|
Initialize the filter with a function: |
132
|
|
|
>>> is_named = skchem.filters.Filter(lambda m: m.name is not None) |
133
|
|
|
|
134
|
|
|
Filter results can be found with `transform`: |
135
|
|
|
>>> ethane = skchem.Mol.from_smiles('CC', name='ethane') |
136
|
|
|
>>> is_named.transform(ethane) |
137
|
|
|
True |
138
|
|
|
|
139
|
|
|
>>> anonymous = skchem.Mol.from_smiles('c1ccccc1') |
140
|
|
|
>>> is_named.transform(anonymous) |
141
|
|
|
False |
142
|
|
|
|
143
|
|
|
Can take a series or dataframe: |
144
|
|
|
>>> mols = pd.Series({'anonymous': anonymous, 'ethane': ethane}) |
145
|
|
|
>>> is_named.transform(mols) |
146
|
|
|
anonymous False |
147
|
|
|
ethane True |
148
|
|
|
Name: Filter, dtype: bool |
149
|
|
|
|
150
|
|
|
Using `filter` will drop out molecules that fail the test: |
151
|
|
|
>>> is_named.filter(mols) |
152
|
|
|
ethane <Mol: CC> |
153
|
|
|
dtype: object |
154
|
|
|
|
155
|
|
|
Only failed are retained with the `neg` keyword argument: |
156
|
|
|
>>> is_named.filter(mols, neg=True) |
157
|
|
|
anonymous <Mol: c1ccccc1> |
158
|
|
|
dtype: object |
159
|
|
|
""" |
160
|
1 |
|
def __init__(self, func=None, agg='any', n_jobs=1, verbose=True): |
161
|
|
|
|
162
|
|
|
""" Initialize a `Filter` object. |
163
|
|
|
|
164
|
|
|
Args: |
165
|
|
|
func (function: Mol => bool): |
166
|
|
|
The function to use to filter the arguments. |
167
|
|
|
agg (str or function: iterable<bool> => bool): |
168
|
|
|
The aggregation to use in the filter. Can be 'any', 'all', |
169
|
|
|
'not any', 'not all' or a callable, for example `any` or `all`. |
170
|
|
|
""" |
171
|
|
|
|
172
|
1 |
|
super(Filter, self).__init__(agg=agg, n_jobs=n_jobs, verbose=verbose) |
173
|
1 |
|
if func is not None: |
174
|
1 |
|
self._transform_mol = func |
175
|
|
|
|
176
|
1 |
|
def _transform_mol(self, mol): |
|
|
|
|
177
|
|
|
raise NotImplemented |
|
|
|
|
178
|
|
|
|
179
|
|
|
|
180
|
1 |
|
class TransformFilter(BaseFilter): |
181
|
|
|
|
182
|
|
|
""" Transform Filter object. |
183
|
|
|
|
184
|
|
|
Implements `transform_filter`, which allows a transform, then a |
185
|
|
|
filter step returning the transformed values that are not `False`, `None` |
186
|
|
|
or `np.nan`. |
187
|
|
|
|
188
|
|
|
""" |
189
|
|
|
|
190
|
1 |
|
def transform_filter(self, mols, y=None, neg=False): |
|
|
|
|
191
|
|
|
|
192
|
|
|
res = self.transform(mols) |
193
|
|
|
mask = self._mask(res=res, neg=neg) |
194
|
|
|
|
195
|
|
|
if isinstance(mols, core.Mol): |
196
|
|
|
return res if mask else None |
197
|
|
|
|
198
|
|
|
if y is None: |
199
|
|
|
return res[mask] |
200
|
|
|
else: |
201
|
|
|
return res[mask], y[mask] |
202
|
|
|
|
This can be caused by one of the following:
1. Missing Dependencies
This error could indicate a configuration issue of Pylint. Make sure that your libraries are available by adding the necessary commands.
2. Missing __init__.py files
This error could also result from missing
__init__.py
files in your module folders. Make sure that you place one file in each sub-folder.