1
|
|
|
""" |
2
|
|
|
Run searches and write files. |
3
|
|
|
""" |
4
|
|
|
|
5
|
|
|
from __future__ import annotations |
6
|
|
|
|
7
|
|
|
import abc |
8
|
|
|
import logging |
9
|
|
|
from inspect import cleandoc as doc |
10
|
|
|
from pathlib import Path |
11
|
|
|
from typing import TypeVar, Generic, Union, Mapping, Set, Sequence, Type, Optional |
12
|
|
|
|
13
|
|
|
import typer |
|
|
|
|
14
|
|
|
|
15
|
|
|
from mandos.model import ReflectionUtils, InjectionError |
16
|
|
|
from mandos.model.chembl_api import ChemblApi |
17
|
|
|
from mandos.model.chembl_support import DataValidityComment |
18
|
|
|
from mandos.model.chembl_support.chembl_targets import TargetType, ConfidenceLevel |
19
|
|
|
from mandos.model.pubchem_support.pubchem_models import ( |
20
|
|
|
ClinicalTrialsGovUtils, |
21
|
|
|
CoOccurrenceType, |
22
|
|
|
AssayType, |
23
|
|
|
) |
24
|
|
|
from mandos.model.searches import Search |
25
|
|
|
from mandos.model.settings import MANDOS_SETTINGS |
26
|
|
|
from mandos.model.taxonomy import Taxonomy |
27
|
|
|
from mandos.model.taxonomy_caches import TaxonomyFactories |
28
|
|
|
from mandos.entries.api_singletons import Apis |
29
|
|
|
from mandos.search.chembl.target_traversal import TargetTraversalStrategies |
30
|
|
|
from mandos.search.pubchem.bioactivity_search import BioactivitySearch |
31
|
|
|
from mandos.search.pubchem.dgidb_search import DgiSearch |
32
|
|
|
from mandos.search.pubchem.ctd_gene_search import CtdGeneSearch |
33
|
|
|
from mandos.entries.searcher import Searcher |
34
|
|
|
from mandos.search.pubchem.drugbank_ddi_search import DrugbankDdiSearch |
35
|
|
|
from mandos.search.pubchem.drugbank_interaction_search import ( |
36
|
|
|
DrugbankTargetSearch, |
37
|
|
|
DrugbankGeneralFunctionSearch, |
38
|
|
|
) |
39
|
|
|
|
40
|
|
|
Chembl, Pubchem = Apis.Chembl, Apis.Pubchem |
41
|
|
|
from mandos.search.chembl.binding_search import BindingSearch |
|
|
|
|
42
|
|
|
from mandos.search.chembl.atc_search import AtcSearch |
|
|
|
|
43
|
|
|
from mandos.search.chembl.go_search import GoType, GoSearch |
|
|
|
|
44
|
|
|
from mandos.search.chembl.indication_search import IndicationSearch |
|
|
|
|
45
|
|
|
from mandos.search.chembl.mechanism_search import MechanismSearch |
|
|
|
|
46
|
|
|
from mandos.search.pubchem.cooccurrence_search import ( |
|
|
|
|
47
|
|
|
GeneCoOccurrenceSearch, |
48
|
|
|
ChemicalCoOccurrenceSearch, |
49
|
|
|
CoOccurrenceSearch, |
50
|
|
|
) |
51
|
|
|
from mandos.search.pubchem.disease_search import DiseaseSearch |
|
|
|
|
52
|
|
|
|
53
|
|
|
logger = logging.getLogger(__package__) |
54
|
|
|
|
55
|
|
|
S = TypeVar("S", bound=Search, covariant=True) |
|
|
|
|
56
|
|
|
U = TypeVar("U", covariant=True, bound=CoOccurrenceSearch) |
|
|
|
|
57
|
|
|
|
58
|
|
|
|
59
|
|
|
class Utils: |
|
|
|
|
60
|
|
|
"""""" |
61
|
|
|
|
62
|
|
|
@staticmethod |
63
|
|
|
def split(st: str) -> Set[str]: |
|
|
|
|
64
|
|
|
return {s.strip() for s in st.split(",")} |
65
|
|
|
|
66
|
|
|
@staticmethod |
67
|
|
|
def get_taxa(taxa: str) -> Sequence[Taxonomy]: |
|
|
|
|
68
|
|
|
return [ |
69
|
|
|
TaxonomyFactories.from_uniprot(MANDOS_SETTINGS.taxonomy_cache_path).load(int(taxon)) |
70
|
|
|
for taxon in taxa.split(",") |
71
|
|
|
] |
72
|
|
|
|
73
|
|
|
@staticmethod |
74
|
|
|
def get_trial_statuses(st: str) -> Set[str]: |
|
|
|
|
75
|
|
|
return ClinicalTrialsGovUtils.resolve_statuses(st) |
76
|
|
|
|
77
|
|
|
@staticmethod |
78
|
|
|
def get_target_types(st: str) -> Set[str]: |
|
|
|
|
79
|
|
|
return {s.name for s in TargetType.resolve(st)} |
80
|
|
|
|
81
|
|
|
@staticmethod |
82
|
|
|
def get_flags(st: str) -> Set[str]: |
|
|
|
|
83
|
|
|
return {s.name for s in DataValidityComment.resolve(st)} |
84
|
|
|
|
85
|
|
|
|
86
|
|
|
class _Typer: |
87
|
|
|
|
88
|
|
|
path = typer.Argument( |
89
|
|
|
None, |
90
|
|
|
exists=True, |
91
|
|
|
dir_okay=False, |
92
|
|
|
readable=True, |
93
|
|
|
help=doc( |
94
|
|
|
""" |
95
|
|
|
The path to the input file. |
96
|
|
|
One of: |
97
|
|
|
|
98
|
|
|
(A) *.txt or *.lines with one InChI Key per line; |
99
|
|
|
|
100
|
|
|
(B) A *.csv, *.tsv, *.tab file (or .gzip variant) with a column called 'inchikey'; OR |
101
|
|
|
|
102
|
|
|
(C) An Apache Arrow *.feather file with a column called 'inchikey' |
103
|
|
|
""" |
104
|
|
|
), |
105
|
|
|
) |
106
|
|
|
|
107
|
|
|
@staticmethod |
108
|
|
|
def key(name: str) -> typer.Option: |
|
|
|
|
109
|
|
|
return typer.Option( |
110
|
|
|
name, |
111
|
|
|
min=1, |
112
|
|
|
max=120, |
113
|
|
|
help=""" |
114
|
|
|
A free-text unique key for the search. |
115
|
|
|
Should be a short, <60-character name that describes the search and any parameters. |
116
|
|
|
The output file will be named according to a 'sanitized' variant of this value. |
117
|
|
|
""", |
118
|
|
|
) |
119
|
|
|
|
120
|
|
|
taxa = typer.Option( |
121
|
|
|
"7742", |
122
|
|
|
show_default=False, |
123
|
|
|
help=doc( |
124
|
|
|
""" |
125
|
|
|
The IDs or names of UniProt taxa, comma-separated. |
126
|
|
|
Taxon names and common names can be used for vertebrate species (where available). |
127
|
|
|
|
128
|
|
|
This can have a significant effect on the search. See the docs fore more info. |
129
|
|
|
|
130
|
|
|
[default: 7742] (Euteleostomi) |
131
|
|
|
""" |
132
|
|
|
), |
133
|
|
|
) |
134
|
|
|
|
135
|
|
|
traversal_strategy = typer.Option( |
136
|
|
|
"@null", |
137
|
|
|
show_default=False, |
138
|
|
|
help=doc( |
139
|
|
|
""" |
140
|
|
|
Target traversal strategy name, file, or class. |
141
|
|
|
Dictates the way the network of ChEMBL targets is traversed (from the annotated target as a source). |
|
|
|
|
142
|
|
|
Specifies the network links that are followed and which targets are 'accepted' for final annotations. |
|
|
|
|
143
|
|
|
This option has a dramatic effect on the search. See the docs for more info. |
144
|
|
|
|
145
|
|
|
Can be one of: |
146
|
|
|
(A) A standard strategy name, starting with @; |
147
|
|
|
(B) The path to a ``*.strat`` file; OR |
148
|
|
|
(C) The fully qualified name of a ``TargetTraversal`` |
149
|
|
|
|
150
|
|
|
The standard traversal strategies are: {} |
151
|
|
|
|
152
|
|
|
[default: @null] (No traversal; targets as-is) |
153
|
|
|
""".format( |
154
|
|
|
"; ".join(TargetTraversalStrategies.standard_strategies()) |
|
|
|
|
155
|
|
|
) |
156
|
|
|
), |
157
|
|
|
) |
158
|
|
|
|
159
|
|
|
target_types = typer.Option( |
160
|
|
|
"@molecular", |
161
|
|
|
show_default=False, |
162
|
|
|
help=doc( |
163
|
|
|
""" |
164
|
|
|
The accepted target types, comma-separated. |
165
|
|
|
|
166
|
|
|
NOTE: This affects only the types are are accepted after traversal, |
167
|
|
|
and the types must be included in the traversal. |
168
|
|
|
This means that this must be AT LEAST as restrictive as the traversal strategy. |
169
|
|
|
|
170
|
|
|
The ChEMBL-defined types are: |
171
|
|
|
|
172
|
|
|
{} |
173
|
|
|
|
174
|
|
|
These special names are also accepted: |
175
|
|
|
|
176
|
|
|
- {} |
177
|
|
|
|
178
|
|
|
[default: @molecular] |
179
|
|
|
""".format( |
180
|
|
|
"; ".join([s.name for s in TargetType.all_types()]), |
|
|
|
|
181
|
|
|
"\n\n - ".join( |
|
|
|
|
182
|
|
|
[f"{k} ({v})" for k, v in TargetType.special_type_names().items()] |
183
|
|
|
), |
184
|
|
|
) |
185
|
|
|
), |
186
|
|
|
) |
187
|
|
|
|
188
|
|
|
min_confidence = typer.Option( |
189
|
|
|
3, |
190
|
|
|
min=0, |
191
|
|
|
max=9, |
192
|
|
|
show_default=False, |
193
|
|
|
help=doc( |
194
|
|
|
""" |
195
|
|
|
Minimum target confidence score, inclusive. |
196
|
|
|
This is useful to modify in only some cases. More important options are min_pchembl and taxa. |
|
|
|
|
197
|
|
|
|
198
|
|
|
Values are: {} |
199
|
|
|
|
200
|
|
|
[default: 3] |
201
|
|
|
""".format( |
202
|
|
|
"; ".join([f"{s.value} ({s.name})" for s in ConfidenceLevel]) |
|
|
|
|
203
|
|
|
) |
204
|
|
|
), |
205
|
|
|
) |
206
|
|
|
|
207
|
|
|
relations = typer.Option( |
208
|
|
|
"<,<=,=", |
209
|
|
|
show_default=False, |
210
|
|
|
help=doc( |
211
|
|
|
""" |
212
|
|
|
Assay activity relations allowed, comma-separated. |
213
|
|
|
If post-processing yourself, consider including all. |
214
|
|
|
Values are: <, <=, =, >, >=, ~. |
215
|
|
|
[default: <,<=,=] |
216
|
|
|
""" |
217
|
|
|
), |
218
|
|
|
) |
219
|
|
|
|
220
|
|
|
min_pchembl = typer.Option( |
221
|
|
|
6.0, |
222
|
|
|
min=0.0, |
223
|
|
|
show_default=False, |
224
|
|
|
help=doc( |
225
|
|
|
""" |
226
|
|
|
Minimum pCHEMBL value, inclusive. |
227
|
|
|
If post-processing yourself, consider setting to 0.0. |
228
|
|
|
[default: 6.0] |
229
|
|
|
""" |
230
|
|
|
), |
231
|
|
|
) |
232
|
|
|
|
233
|
|
|
banned_flags = typer.Option( |
234
|
|
|
"@negative", |
235
|
|
|
show_default=False, |
236
|
|
|
help=doc( |
237
|
|
|
""" |
238
|
|
|
Exclude activity annotations with data validity flags, comma-separated. |
239
|
|
|
It is rare to need to change this. |
240
|
|
|
|
241
|
|
|
Values are: {}. |
242
|
|
|
|
243
|
|
|
Special sets are: |
244
|
|
|
|
245
|
|
|
- @all (all flags are banned) |
246
|
|
|
|
247
|
|
|
- @negative ({}) |
248
|
|
|
|
249
|
|
|
- @positive ({}) |
250
|
|
|
|
251
|
|
|
[default: @negative] |
252
|
|
|
""".format( |
253
|
|
|
"; ".join([s.name for s in DataValidityComment]), |
|
|
|
|
254
|
|
|
", ".join([s.name for s in DataValidityComment.negative_comments()]), |
|
|
|
|
255
|
|
|
", ".join([s.name for s in DataValidityComment.positive_comments()]), |
|
|
|
|
256
|
|
|
), |
257
|
|
|
), |
258
|
|
|
) |
259
|
|
|
|
260
|
|
|
test = typer.Option(False, help="Do not run searches; just check that the parameters are ok.") |
261
|
|
|
|
262
|
|
|
chembl_trial = typer.Option( |
263
|
|
|
3, |
264
|
|
|
show_default=False, |
265
|
|
|
help=doc( |
266
|
|
|
""" |
267
|
|
|
Minimum phase of a clinical trial, inclusive. |
268
|
|
|
Values are: 0, 1, 2, 3. |
269
|
|
|
[default: 3] |
270
|
|
|
""" |
271
|
|
|
), |
272
|
|
|
min=0, |
273
|
|
|
max=3, |
274
|
|
|
) |
275
|
|
|
|
276
|
|
|
|
277
|
|
|
class Entry(Generic[S], metaclass=abc.ABCMeta): |
|
|
|
|
278
|
|
|
@classmethod |
279
|
|
|
def cmd(cls) -> str: |
|
|
|
|
280
|
|
|
key = cls._get_default_key() |
281
|
|
|
if isinstance(key, typer.models.OptionInfo): |
282
|
|
|
key = key.default |
283
|
|
|
if key is None or not isinstance(key, str): |
284
|
|
|
raise AssertionError(f"Key for {cls.__name__} is {key}") |
285
|
|
|
return key |
286
|
|
|
|
287
|
|
|
@classmethod |
288
|
|
|
def run(cls, path: Path, **params) -> None: |
|
|
|
|
289
|
|
|
raise NotImplementedError() |
290
|
|
|
|
291
|
|
|
@classmethod |
292
|
|
|
def get_search_type(cls) -> Type[S]: |
|
|
|
|
293
|
|
|
# noinspection PyTypeChecker |
294
|
|
|
return ReflectionUtils.get_generic_arg(cls, Search) |
295
|
|
|
|
296
|
|
|
@classmethod |
297
|
|
|
def test(cls, path: Path, **params) -> None: |
|
|
|
|
298
|
|
|
params = dict(params) |
299
|
|
|
params["test"] = True |
300
|
|
|
cls.run(**params) |
301
|
|
|
|
302
|
|
|
@classmethod |
303
|
|
|
def _run(cls, built: S, path: Path, test: bool): |
304
|
|
|
searcher = Searcher([built], path) |
305
|
|
|
if not test: |
306
|
|
|
searcher.search() |
307
|
|
|
return searcher |
308
|
|
|
|
309
|
|
|
# @classmethod |
310
|
|
|
# def build(cls, path: Path, **params: Mapping[str, Union[int, float, str]]) -> Search: |
311
|
|
|
# raise NotImplementedError() |
312
|
|
|
|
313
|
|
|
@classmethod |
314
|
|
|
def default_param_values(cls) -> Mapping[str, Union[str, float, int]]: |
|
|
|
|
315
|
|
|
return { |
316
|
|
|
param: value |
317
|
|
|
for param, value in ReflectionUtils.default_arg_values(cls.run).items() |
318
|
|
|
if param not in {"key", "path"} |
319
|
|
|
} |
320
|
|
|
|
321
|
|
|
@classmethod |
322
|
|
|
def _get_default_key(cls) -> str: |
323
|
|
|
vals = ReflectionUtils.default_arg_values(cls.run) |
324
|
|
|
try: |
325
|
|
|
return vals["key"] |
326
|
|
|
except KeyError: |
327
|
|
|
logger.error(f"key not in {vals.keys()} for {cls.__name__}") |
|
|
|
|
328
|
|
|
raise |
329
|
|
|
|
330
|
|
|
|
331
|
|
|
class EntryChemblBinding(Entry[BindingSearch]): |
|
|
|
|
332
|
|
|
@classmethod |
333
|
|
|
def run( |
|
|
|
|
334
|
|
|
cls, |
|
|
|
|
335
|
|
|
path: Path = _Typer.path, |
|
|
|
|
336
|
|
|
key: str = _Typer.key("chembl:binding"), |
|
|
|
|
337
|
|
|
taxa: Optional[str] = _Typer.taxa, |
|
|
|
|
338
|
|
|
traversal=_Typer.traversal_strategy, |
|
|
|
|
339
|
|
|
target_types=_Typer.target_types, |
|
|
|
|
340
|
|
|
confidence=_Typer.min_confidence, |
|
|
|
|
341
|
|
|
relations=_Typer.relations, |
|
|
|
|
342
|
|
|
min_pchembl=_Typer.min_pchembl, |
|
|
|
|
343
|
|
|
banned_flags=_Typer.banned_flags, |
|
|
|
|
344
|
|
|
test: bool = _Typer.test, |
|
|
|
|
345
|
|
|
) -> Searcher: |
346
|
|
|
""" |
347
|
|
|
Binding data from ChEMBL. |
348
|
|
|
These are 'activity' annotations of the type 'B' that have a pCHEMBL value. |
349
|
|
|
There is extended documentation on this search; see: |
350
|
|
|
|
351
|
|
|
https://mandos-chem.readthedocs.io/en/latest/binding.html |
352
|
|
|
|
353
|
|
|
OBJECT: ChEMBL preferred target name |
354
|
|
|
|
355
|
|
|
PREDICATE: "binds" |
356
|
|
|
""" |
357
|
|
|
built = BindingSearch( |
358
|
|
|
key=key, |
359
|
|
|
api=Chembl, |
360
|
|
|
taxa=Utils.get_taxa(taxa), |
361
|
|
|
traversal_strategy=traversal, |
362
|
|
|
allowed_target_types=Utils.get_target_types(target_types), |
363
|
|
|
min_confidence_score=confidence, |
364
|
|
|
allowed_relations=Utils.split(relations), |
365
|
|
|
min_pchembl=min_pchembl, |
366
|
|
|
banned_flags=Utils.get_flags(banned_flags), |
367
|
|
|
) |
368
|
|
|
return cls._run(built, path, test) |
369
|
|
|
|
370
|
|
|
|
371
|
|
|
class EntryChemblMechanism(Entry[MechanismSearch]): |
|
|
|
|
372
|
|
|
@classmethod |
373
|
|
|
def run( |
|
|
|
|
374
|
|
|
cls, |
|
|
|
|
375
|
|
|
path: Path = _Typer.path, |
|
|
|
|
376
|
|
|
key: str = _Typer.key("chembl:mechanism"), |
|
|
|
|
377
|
|
|
taxa: Optional[str] = _Typer.taxa, |
|
|
|
|
378
|
|
|
traversal: str = _Typer.traversal_strategy, |
|
|
|
|
379
|
|
|
target_types: str = _Typer.target_types, |
|
|
|
|
380
|
|
|
min_confidence: Optional[int] = _Typer.min_confidence, |
|
|
|
|
381
|
|
|
test: bool = _Typer.test, |
|
|
|
|
382
|
|
|
) -> Searcher: |
383
|
|
|
""" |
384
|
|
|
Mechanism of action (MoA) data from ChEMBL. |
385
|
|
|
|
386
|
|
|
OBJECT: ChEMBL preferred target name |
387
|
|
|
|
388
|
|
|
PREDICATE: Target action; e.g. "agonist" or "positive allosteric modulator" |
389
|
|
|
""" |
390
|
|
|
built = MechanismSearch( |
391
|
|
|
key=key, |
392
|
|
|
api=Chembl, |
393
|
|
|
taxa=Utils.get_taxa(taxa), |
394
|
|
|
traversal_strategy=traversal, |
395
|
|
|
allowed_target_types=Utils.get_target_types(target_types), |
396
|
|
|
min_confidence_score=min_confidence, |
397
|
|
|
) |
398
|
|
|
return cls._run(built, path, test) |
399
|
|
|
|
400
|
|
|
|
401
|
|
|
class EntryChemblTrials(Entry[IndicationSearch]): |
|
|
|
|
402
|
|
|
@classmethod |
403
|
|
|
def run( |
404
|
|
|
cls, |
|
|
|
|
405
|
|
|
path: Path = _Typer.path, |
|
|
|
|
406
|
|
|
key: str = _Typer.key("chembl.trial"), |
|
|
|
|
407
|
|
|
min_phase: Optional[int] = _Typer.chembl_trial, |
|
|
|
|
408
|
|
|
test: bool = _Typer.test, |
|
|
|
|
409
|
|
|
) -> Searcher: |
410
|
|
|
""" |
411
|
|
|
Diseases from clinical trials listed in ChEMBL. |
412
|
|
|
|
413
|
|
|
OBJECT: MeSH code |
414
|
|
|
|
415
|
|
|
PREDICATE: "phase <level> trial" |
416
|
|
|
""" |
417
|
|
|
built = IndicationSearch(key=key, api=Chembl, min_phase=min_phase) |
418
|
|
|
return cls._run(built, path, test) |
419
|
|
|
|
420
|
|
|
|
421
|
|
|
class EntryChemblAtc(Entry[AtcSearch]): |
|
|
|
|
422
|
|
|
@classmethod |
423
|
|
|
def run( |
424
|
|
|
cls, |
|
|
|
|
425
|
|
|
path: Path = _Typer.path, |
|
|
|
|
426
|
|
|
key: str = _Typer.key("chembl.atc"), |
|
|
|
|
427
|
|
|
levels: str = typer.Option( |
|
|
|
|
428
|
|
|
"1,2,3,4,5", min=1, max=5, help="""List of ATC levels, comma-separated.""" |
429
|
|
|
), |
430
|
|
|
test: bool = _Typer.test, |
|
|
|
|
431
|
|
|
) -> Searcher: |
432
|
|
|
""" |
433
|
|
|
ATC codes from ChEMBL. |
434
|
|
|
|
435
|
|
|
OBJECT: ATC Code |
436
|
|
|
|
437
|
|
|
PREDICATE: "ATC L<leveL> code" |
438
|
|
|
""" |
439
|
|
|
built = AtcSearch(key=key, api=Chembl, levels={int(x.strip()) for x in levels.split(",")}) |
440
|
|
|
return cls._run(built, path, test) |
441
|
|
|
|
442
|
|
|
|
443
|
|
|
class _EntryChemblGo(Entry[GoSearch], metaclass=abc.ABCMeta): |
444
|
|
|
@classmethod |
445
|
|
|
def go_type(cls) -> GoType: |
|
|
|
|
446
|
|
|
raise NotImplementedError() |
447
|
|
|
|
448
|
|
|
@classmethod |
449
|
|
|
def cmd(cls) -> str: |
|
|
|
|
450
|
|
|
return f"chembl:go.{cls.go_type().name.lower()}" |
451
|
|
|
|
452
|
|
|
@classmethod |
453
|
|
|
def run( |
|
|
|
|
454
|
|
|
cls, |
|
|
|
|
455
|
|
|
path: Path = _Typer.path, |
|
|
|
|
456
|
|
|
key: str = _Typer.key("<see above>"), |
|
|
|
|
457
|
|
|
taxa: Optional[str] = _Typer.taxa, |
|
|
|
|
458
|
|
|
traversal_strategy: str = _Typer.traversal_strategy, |
|
|
|
|
459
|
|
|
target_types: str = _Typer.target_types, |
|
|
|
|
460
|
|
|
confidence: Optional[int] = _Typer.min_confidence, |
|
|
|
|
461
|
|
|
relations: str = _Typer.relations, |
|
|
|
|
462
|
|
|
min_pchembl: float = _Typer.min_pchembl, |
|
|
|
|
463
|
|
|
banned_flags: Optional[str] = _Typer.banned_flags, |
|
|
|
|
464
|
|
|
binding_search: Optional[str] = typer.Option( |
|
|
|
|
465
|
|
|
None, |
466
|
|
|
help=""" |
467
|
|
|
The fully qualified name of a class inheriting ``BindingSearch``. |
468
|
|
|
If specified, all parameters above are passed to its constructor. |
469
|
|
|
""", |
470
|
|
|
), |
471
|
|
|
test=_Typer.test, |
|
|
|
|
472
|
|
|
) -> Searcher: |
473
|
|
|
""" |
474
|
|
|
GO terms associated with ChEMBL binding targets. |
475
|
|
|
|
476
|
|
|
OBJECT: GO Term name |
477
|
|
|
|
478
|
|
|
PREDICATE: "GO <Function/Process/Component> term" |
479
|
|
|
|
480
|
|
|
Note: |
481
|
|
|
|
482
|
|
|
By default, the key is the "chembl:go.function", "chembl:go.process", or "chembl:go.component". |
|
|
|
|
483
|
|
|
""" |
484
|
|
|
if key is None: |
485
|
|
|
key = cls.cmd() |
486
|
|
|
api = ChemblApi.wrap(Chembl) |
487
|
|
|
if binding_search is None: |
488
|
|
|
binding_clazz = BindingSearch |
489
|
|
|
else: |
490
|
|
|
binding_clazz = ReflectionUtils.injection(binding_search, BindingSearch) |
491
|
|
|
logger.info(f"NOTICE: Passing parameters to {binding_clazz.__qualname__}") |
|
|
|
|
492
|
|
|
try: |
493
|
|
|
binding_search = binding_clazz( |
494
|
|
|
key=key, |
495
|
|
|
api=Chembl, |
496
|
|
|
taxa=Utils.get_taxa(taxa), |
497
|
|
|
traversal_strategy=traversal_strategy, |
498
|
|
|
allowed_target_types=Utils.get_target_types(target_types), |
499
|
|
|
min_confidence_score=confidence, |
500
|
|
|
allowed_relations=Utils.split(relations), |
501
|
|
|
min_pchembl=min_pchembl, |
502
|
|
|
banned_flags=Utils.get_flags(banned_flags), |
503
|
|
|
) |
504
|
|
|
except (TypeError, ValueError): |
505
|
|
|
raise InjectionError(f"Failed to build {binding_clazz.__qualname__}") |
506
|
|
|
built = GoSearch(key, api, cls.go_type(), binding_search) |
507
|
|
|
return cls._run(built, path, test) |
508
|
|
|
|
509
|
|
|
|
510
|
|
|
class EntryGoFunction(_EntryChemblGo): |
|
|
|
|
511
|
|
|
@classmethod |
512
|
|
|
def go_type(cls) -> GoType: |
513
|
|
|
return GoType.function |
514
|
|
|
|
515
|
|
|
|
516
|
|
|
class EntryGoProcess(_EntryChemblGo): |
|
|
|
|
517
|
|
|
@classmethod |
518
|
|
|
def go_type(cls) -> GoType: |
519
|
|
|
return GoType.process |
520
|
|
|
|
521
|
|
|
|
522
|
|
|
class EntryGoComponent(_EntryChemblGo): |
|
|
|
|
523
|
|
|
@classmethod |
524
|
|
|
def go_type(cls) -> GoType: |
525
|
|
|
return GoType.component |
526
|
|
|
|
527
|
|
|
|
528
|
|
|
class EntryPubchemDisease(Entry[DiseaseSearch]): |
|
|
|
|
529
|
|
|
@classmethod |
530
|
|
|
def run( |
|
|
|
|
531
|
|
|
cls, |
|
|
|
|
532
|
|
|
path: Path = _Typer.path, |
|
|
|
|
533
|
|
|
key: str = _Typer.key("disease.ctd:mesh"), |
|
|
|
|
534
|
|
|
therapeutic: bool = typer.Option(True, help="Include annotations of type 'therapeutic'"), |
|
|
|
|
535
|
|
|
marker: bool = typer.Option(True, help="Include annotations of type 'marker/mechanism'"), |
|
|
|
|
536
|
|
|
test: bool = _Typer.test, |
|
|
|
|
537
|
|
|
) -> Searcher: |
538
|
|
|
""" |
539
|
|
|
Diseases in the Comparative Toxicogenomics Database (CTD). |
540
|
|
|
|
541
|
|
|
OBJECT: MeSH code of disease |
542
|
|
|
|
543
|
|
|
PREDICATE: "marker/mechanism" or "disease" |
544
|
|
|
""" |
545
|
|
|
built = DiseaseSearch(key, Pubchem, therapeutic=therapeutic, marker=marker) |
546
|
|
|
return cls._run(built, path, test) |
547
|
|
|
|
548
|
|
|
|
549
|
|
|
class _EntryPubchemCoOccurrence(Entry[U], metaclass=abc.ABCMeta): |
550
|
|
|
@classmethod |
551
|
|
|
def cmd(cls) -> str: |
|
|
|
|
552
|
|
|
return f"lit.pubchem:{cls.get_cooccurrence_type().name.lower()}" |
553
|
|
|
|
554
|
|
|
@classmethod |
555
|
|
|
def get_cooccurrence_type(cls) -> CoOccurrenceType: |
|
|
|
|
556
|
|
|
s: CoOccurrenceSearch = cls.get_search_type() |
|
|
|
|
557
|
|
|
return s.cooccurrence_type() |
558
|
|
|
|
559
|
|
|
@classmethod |
560
|
|
|
def run( |
|
|
|
|
561
|
|
|
cls, |
|
|
|
|
562
|
|
|
path: Path = _Typer.path, |
|
|
|
|
563
|
|
|
key: str = _Typer.key("<see above>"), |
|
|
|
|
564
|
|
|
min_score: float = typer.Option( |
|
|
|
|
565
|
|
|
0.0, |
566
|
|
|
help="Minimum enrichment score, inclusive. See docs for more info.", |
567
|
|
|
min=0.0, |
568
|
|
|
), |
569
|
|
|
min_articles: int = typer.Option( |
|
|
|
|
570
|
|
|
0, |
571
|
|
|
help="Minimum number of articles for both the compound and object, inclusive.", |
572
|
|
|
min=0, |
573
|
|
|
), |
574
|
|
|
test: bool = _Typer.test, |
|
|
|
|
575
|
|
|
) -> Searcher: |
576
|
|
|
""" |
577
|
|
|
Co-occurrences from PubMed articles. |
578
|
|
|
There is extended documentation on this search. |
579
|
|
|
Also refer to https://pubchemdocs.ncbi.nlm.nih.gov/knowledge-panels |
580
|
|
|
|
581
|
|
|
OBJECT: Name of gene/chemical/disease |
582
|
|
|
|
583
|
|
|
PREDICATE: "<gene/chemical/disease> co-occurrence" |
584
|
|
|
""" |
585
|
|
|
if key is None: |
586
|
|
|
key = cls.cmd() |
587
|
|
|
clazz = cls.get_search_type() |
588
|
|
|
built = clazz(key, Pubchem, min_score=min_score, min_articles=min_articles) |
589
|
|
|
return cls._run(built, path, test) |
590
|
|
|
|
591
|
|
|
|
592
|
|
|
class EntryPubchemGeneCoOccurrence(_EntryPubchemCoOccurrence[GeneCoOccurrenceSearch]): |
|
|
|
|
593
|
|
|
"""""" |
594
|
|
|
|
595
|
|
|
|
596
|
|
|
class EntryPubchemDiseaseCoOccurrence(_EntryPubchemCoOccurrence[GeneCoOccurrenceSearch]): |
|
|
|
|
597
|
|
|
"""""" |
598
|
|
|
|
599
|
|
|
|
600
|
|
|
class EntryPubchemChemicalCoOccurrence(_EntryPubchemCoOccurrence[ChemicalCoOccurrenceSearch]): |
|
|
|
|
601
|
|
|
"""""" |
602
|
|
|
|
603
|
|
|
|
604
|
|
|
class EntryPubchemDgi(Entry[DgiSearch]): |
|
|
|
|
605
|
|
|
@classmethod |
606
|
|
|
def run( |
607
|
|
|
cls, |
|
|
|
|
608
|
|
|
path: Path = _Typer.path, |
|
|
|
|
609
|
|
|
key: str = _Typer.key("interact.dgidb:gene"), |
|
|
|
|
610
|
|
|
test: bool = _Typer.test, |
|
|
|
|
611
|
|
|
) -> Searcher: |
612
|
|
|
""" |
613
|
|
|
Drug/gene interactions in the Drug Gene Interaction Database (DGIDB). |
614
|
|
|
Also see ``disease.dgidb:int``. |
615
|
|
|
|
616
|
|
|
OBJECT: Name of the gene |
617
|
|
|
|
618
|
|
|
PREDICATE: "drug/gene interaction" |
619
|
|
|
""" |
620
|
|
|
built = DgiSearch(key, Pubchem) |
621
|
|
|
return cls._run(built, path, test) |
622
|
|
|
|
623
|
|
|
|
624
|
|
|
class EntryPubchemCgi(Entry[CtdGeneSearch]): |
|
|
|
|
625
|
|
|
@classmethod |
626
|
|
|
def run( |
627
|
|
|
cls, |
|
|
|
|
628
|
|
|
path: Path = _Typer.path, |
|
|
|
|
629
|
|
|
key: str = _Typer.key("interact.ctd:gene"), |
|
|
|
|
630
|
|
|
test: bool = _Typer.test, |
|
|
|
|
631
|
|
|
) -> Searcher: |
632
|
|
|
""" |
633
|
|
|
Compound/gene interactions in the Drug Gene Interaction Database (DGIDB). |
634
|
|
|
Also see ``interact.dgidb:int``. |
635
|
|
|
|
636
|
|
|
OBJECT: Name of the gene |
637
|
|
|
|
638
|
|
|
PREDICATE: "compound/gene interaction" |
639
|
|
|
|
640
|
|
|
""" |
641
|
|
|
built = CtdGeneSearch(key, Pubchem) |
642
|
|
|
return cls._run(built, path, test) |
643
|
|
|
|
644
|
|
|
|
645
|
|
|
class EntryDrugbankTarget(Entry[DrugbankTargetSearch]): |
|
|
|
|
646
|
|
|
@classmethod |
647
|
|
|
def run( |
648
|
|
|
cls, |
|
|
|
|
649
|
|
|
path: Path = _Typer.path, |
|
|
|
|
650
|
|
|
key: str = _Typer.key("interact.drugbank:target"), |
|
|
|
|
651
|
|
|
test: bool = _Typer.test, |
|
|
|
|
652
|
|
|
) -> Searcher: |
653
|
|
|
""" |
654
|
|
|
Protein targets from DrugBank. |
655
|
|
|
|
656
|
|
|
OBJECT: Target name (e.g. "Solute carrier family 22 member 11") from DrugBank |
657
|
|
|
|
658
|
|
|
PREDICATE: Action (e.g. "binder", "downregulator", or "agonist") |
659
|
|
|
""" |
660
|
|
|
built = DrugbankTargetSearch(key, Pubchem) |
661
|
|
|
return cls._run(built, path, test) |
662
|
|
|
|
663
|
|
|
|
664
|
|
|
class EntryGeneralFunction(Entry[DrugbankGeneralFunctionSearch]): |
|
|
|
|
665
|
|
|
@classmethod |
666
|
|
|
def run( |
667
|
|
|
cls, |
|
|
|
|
668
|
|
|
path: Path = _Typer.path, |
|
|
|
|
669
|
|
|
key: str = _Typer.key("interact.drugbank:function"), |
|
|
|
|
670
|
|
|
test: bool = _Typer.test, |
|
|
|
|
671
|
|
|
) -> Searcher: |
672
|
|
|
""" |
673
|
|
|
General functions from DrugBank targets. |
674
|
|
|
|
675
|
|
|
OBJECT: Name of the general function (e.g. "Toxic substance binding") |
676
|
|
|
|
677
|
|
|
PREDICATE: against on target (e.g. "binder", "downregulator", or "agonist"). |
678
|
|
|
""" |
679
|
|
|
built = DrugbankGeneralFunctionSearch(key, Pubchem) |
680
|
|
|
return cls._run(built, path, test) |
681
|
|
|
|
682
|
|
|
|
683
|
|
|
class EntryDrugbankDdi(Entry[DrugbankDdiSearch]): |
|
|
|
|
684
|
|
|
@classmethod |
685
|
|
|
def run( |
686
|
|
|
cls, |
|
|
|
|
687
|
|
|
path: Path = _Typer.path, |
|
|
|
|
688
|
|
|
key: str = _Typer.key("interact.drugbank:ddi"), |
|
|
|
|
689
|
|
|
test: bool = _Typer.test, |
|
|
|
|
690
|
|
|
) -> Searcher: |
691
|
|
|
""" |
692
|
|
|
Drug/drug interactions listed by DrugBank. |
693
|
|
|
|
694
|
|
|
The 'description' column includes useful information about the interaction, |
695
|
|
|
such as diseases and whether a risk is increased or decreased. |
696
|
|
|
|
697
|
|
|
OBJECT: name of the drug (e.g. "ibuprofen") |
698
|
|
|
|
699
|
|
|
PREDICATE: "ddi" |
700
|
|
|
""" |
701
|
|
|
built = DrugbankDdiSearch(key, Pubchem) |
702
|
|
|
return cls._run(built, path, test) |
703
|
|
|
|
704
|
|
|
|
705
|
|
|
class EntryPubchemAssay(Entry[BioactivitySearch]): |
|
|
|
|
706
|
|
|
@classmethod |
707
|
|
|
def run( |
|
|
|
|
708
|
|
|
cls, |
|
|
|
|
709
|
|
|
path: Path = _Typer.path, |
|
|
|
|
710
|
|
|
key: str = _Typer.key("assay.pubchem:activity"), |
|
|
|
|
711
|
|
|
confirmatory: bool = typer.Option(True, help="Include 'confirmatory' assays"), |
|
|
|
|
712
|
|
|
literature: bool = typer.Option(True, help="Include 'literature' assays"), |
|
|
|
|
713
|
|
|
other: bool = typer.Option(True, help="Include 'other' assays"), |
|
|
|
|
714
|
|
|
name_must_match: bool = typer.Option( |
|
|
|
|
715
|
|
|
False, |
716
|
|
|
help=doc( |
717
|
|
|
""" |
718
|
|
|
Require that the name of the compound(s) exactly matches the compound name on PubChem (case-insensitive) |
|
|
|
|
719
|
|
|
""" |
720
|
|
|
), |
721
|
|
|
), |
722
|
|
|
ban_sources: Optional[str] = None, |
|
|
|
|
723
|
|
|
test: bool = _Typer.test, |
|
|
|
|
724
|
|
|
) -> Searcher: |
725
|
|
|
""" |
726
|
|
|
PubChem bioactivity results. |
727
|
|
|
|
728
|
|
|
Note: The species name, if present, is taken from the target name. |
729
|
|
|
The taxon ID is what was curated in PubChem. |
730
|
|
|
|
731
|
|
|
OBJECT: Name of the target without species suffix (e.g. "Slc6a3 - solute carrier family 6 member 3") |
|
|
|
|
732
|
|
|
|
733
|
|
|
PREDICATE: "active"/"inactive"/"inconclusive"/"undetermined" |
734
|
|
|
""" |
735
|
|
|
assay_types = set() |
736
|
|
|
if literature: |
737
|
|
|
assay_types.add(AssayType.literature) |
738
|
|
|
if confirmatory: |
739
|
|
|
assay_types.add(AssayType.confirmatory) |
740
|
|
|
if other: |
741
|
|
|
assay_types.add(AssayType.other) |
742
|
|
|
built = BioactivitySearch( |
743
|
|
|
key, Pubchem, assay_types=assay_types, compound_name_must_match=name_must_match |
744
|
|
|
) |
745
|
|
|
return cls._run(built, path, test) |
746
|
|
|
|
747
|
|
|
|
748
|
|
|
Entries = [ |
749
|
|
|
EntryChemblBinding, |
750
|
|
|
EntryChemblMechanism, |
751
|
|
|
EntryChemblAtc, |
752
|
|
|
EntryChemblTrials, |
753
|
|
|
EntryGoFunction, |
754
|
|
|
EntryGoProcess, |
755
|
|
|
EntryGoComponent, |
756
|
|
|
EntryPubchemDisease, |
757
|
|
|
EntryPubchemGeneCoOccurrence, |
758
|
|
|
EntryPubchemDiseaseCoOccurrence, |
759
|
|
|
EntryPubchemChemicalCoOccurrence, |
760
|
|
|
EntryPubchemDgi, |
761
|
|
|
EntryPubchemCgi, |
762
|
|
|
EntryDrugbankTarget, |
763
|
|
|
EntryGeneralFunction, |
764
|
|
|
EntryDrugbankDdi, |
765
|
|
|
EntryPubchemAssay, |
766
|
|
|
] |
767
|
|
|
|