Datasets:

Modalities:
Text
Formats:
parquet
Languages:
English
Libraries:
Datasets
pandas
License:
jfries commited on
Commit
3c448eb
·
verified ·
1 Parent(s): 53e5a3d

Delete loading script

Browse files
Files changed (1) hide show
  1. chemprot.py +0 -446
chemprot.py DELETED
@@ -1,446 +0,0 @@
1
- # coding=utf-8
2
- # Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
3
- #
4
- # Licensed under the Apache License, Version 2.0 (the "License");
5
- # you may not use this file except in compliance with the License.
6
- # You may obtain a copy of the License at
7
- #
8
- # http://www.apache.org/licenses/LICENSE-2.0
9
- #
10
- # Unless required by applicable law or agreed to in writing, software
11
- # distributed under the License is distributed on an "AS IS" BASIS,
12
- # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13
- # See the License for the specific language governing permissions and
14
- # limitations under the License.
15
- """
16
- The BioCreative VI Chemical-Protein interaction dataset identifies entities of
17
- chemicals and proteins and their likely relation to one other. Compounds are
18
- generally agonists (activators) or antagonists (inhibitors) of proteins. The
19
- script loads dataset in bigbio schema (using knowledgebase schema: schemas/kb)
20
- AND/OR source (default) schema
21
- """
22
- import os
23
- from typing import Dict, Tuple
24
-
25
- import datasets
26
-
27
- from .bigbiohub import kb_features
28
- from .bigbiohub import BigBioConfig
29
- from .bigbiohub import Tasks
30
-
31
- _LANGUAGES = ['English']
32
- _PUBMED = True
33
- _LOCAL = False
34
- _CITATION = """\
35
- @article{DBLP:journals/biodb/LiSJSWLDMWL16,
36
- author = {Krallinger, M., Rabal, O., Lourenço, A.},
37
- title = {Overview of the BioCreative VI chemical-protein interaction Track},
38
- journal = {Proceedings of the BioCreative VI Workshop,},
39
- volume = {141-146},
40
- year = {2017},
41
- url = {https://biocreative.bioinformatics.udel.edu/tasks/biocreative-vi/track-5/},
42
- doi = {},
43
- biburl = {},
44
- bibsource = {}
45
- }
46
- """
47
- _DESCRIPTION = """\
48
- The BioCreative VI Chemical-Protein interaction dataset identifies entities of
49
- chemicals and proteins and their likely relation to one other. Compounds are
50
- generally agonists (activators) or antagonists (inhibitors) of proteins.
51
- """
52
-
53
- _DATASETNAME = "chemprot"
54
- _DISPLAYNAME = "ChemProt"
55
-
56
- _HOMEPAGE = "https://biocreative.bioinformatics.udel.edu/tasks/biocreative-vi/track-5/"
57
-
58
- _LICENSE = 'Public Domain Mark 1.0'
59
-
60
- _URLs = {
61
- "source": "https://huggingface.co/datasets/bigbio/chemprot/resolve/main/ChemProt_Corpus.zip",
62
- "bigbio_kb": "https://huggingface.co/datasets/bigbio/chemprot/resolve/main/ChemProt_Corpus.zip",
63
- }
64
-
65
- _SUPPORTED_TASKS = [Tasks.RELATION_EXTRACTION, Tasks.NAMED_ENTITY_RECOGNITION]
66
- _SOURCE_VERSION = "1.0.0"
67
- _BIGBIO_VERSION = "1.0.0"
68
-
69
-
70
- # Chemprot specific variables
71
- # NOTE: There are 3 examples (2 in dev, 1 in training) with CPR:0
72
- _GROUP_LABELS = {
73
- "CPR:0": "Undefined",
74
- "CPR:1": "Part_of",
75
- "CPR:2": "Regulator",
76
- "CPR:3": "Upregulator",
77
- "CPR:4": "Downregulator",
78
- "CPR:5": "Agonist",
79
- "CPR:6": "Antagonist",
80
- "CPR:7": "Modulator",
81
- "CPR:8": "Cofactor",
82
- "CPR:9": "Substrate",
83
- "CPR:10": "Not",
84
- }
85
-
86
-
87
- class ChemprotDataset(datasets.GeneratorBasedBuilder):
88
- """BioCreative VI Chemical-Protein Interaction Task."""
89
-
90
- SOURCE_VERSION = datasets.Version(_SOURCE_VERSION)
91
- BIGBIO_VERSION = datasets.Version(_BIGBIO_VERSION)
92
-
93
- BUILDER_CONFIGS = [
94
- BigBioConfig(
95
- name="chemprot_full_source",
96
- version=SOURCE_VERSION,
97
- description="chemprot source schema",
98
- schema="source",
99
- subset_id="chemprot_full",
100
- ),
101
- BigBioConfig(
102
- name="chemprot_shared_task_eval_source",
103
- version=SOURCE_VERSION,
104
- description="chemprot source schema with only the relation types that were used in the shared task evaluation",
105
- schema="source",
106
- subset_id="chemprot_shared_task_eval",
107
- ),
108
- BigBioConfig(
109
- name="chemprot_bigbio_kb",
110
- version=BIGBIO_VERSION,
111
- description="chemprot BigBio schema",
112
- schema="bigbio_kb",
113
- subset_id="chemprot",
114
- ),
115
- ]
116
-
117
- DEFAULT_CONFIG_NAME = "chemprot_full_source"
118
-
119
- def _info(self):
120
-
121
- if self.config.schema == "source":
122
- features = datasets.Features(
123
- {
124
- "pmid": datasets.Value("string"),
125
- "text": datasets.Value("string"),
126
- "entities": datasets.Sequence(
127
- {
128
- "id": datasets.Value("string"),
129
- "type": datasets.Value("string"),
130
- "text": datasets.Value("string"),
131
- "offsets": datasets.Sequence(datasets.Value("int64")),
132
- }
133
- ),
134
- "relations": datasets.Sequence(
135
- {
136
- "type": datasets.Value("string"),
137
- "arg1": datasets.Value("string"),
138
- "arg2": datasets.Value("string"),
139
- }
140
- ),
141
- }
142
- )
143
-
144
- elif self.config.schema == "bigbio_kb":
145
- features = kb_features
146
-
147
- return datasets.DatasetInfo(
148
- description=_DESCRIPTION,
149
- features=features,
150
- homepage=_HOMEPAGE,
151
- license=str(_LICENSE),
152
- citation=_CITATION,
153
- )
154
-
155
- def _split_generators(self, dl_manager):
156
- """Returns SplitGenerators."""
157
- my_urls = _URLs[self.config.schema]
158
- data_dir = dl_manager.download_and_extract(my_urls)
159
-
160
- # Extract each of the individual folders
161
- # NOTE: omitting "extract" call cause it uses a new folder
162
- train_path = dl_manager.extract(
163
- os.path.join(data_dir, "ChemProt_Corpus/chemprot_training.zip")
164
- )
165
- test_path = dl_manager.extract(
166
- os.path.join(data_dir, "ChemProt_Corpus/chemprot_test_gs.zip")
167
- )
168
- dev_path = dl_manager.extract(
169
- os.path.join(data_dir, "ChemProt_Corpus/chemprot_development.zip")
170
- )
171
- sample_path = dl_manager.extract(
172
- os.path.join(data_dir, "ChemProt_Corpus/chemprot_sample.zip")
173
- )
174
-
175
- return [
176
- datasets.SplitGenerator(
177
- name="sample", # should be a named split : /
178
- gen_kwargs={
179
- "filepath": os.path.join(sample_path, "chemprot_sample"),
180
- "abstract_file": "chemprot_sample_abstracts.tsv",
181
- "entity_file": "chemprot_sample_entities.tsv",
182
- "relation_file": "chemprot_sample_relations.tsv",
183
- "gold_standard_file": "chemprot_sample_gold_standard.tsv",
184
- "split": "sample",
185
- },
186
- ),
187
- datasets.SplitGenerator(
188
- name=datasets.Split.TRAIN,
189
- gen_kwargs={
190
- "filepath": os.path.join(train_path, "chemprot_training"),
191
- "abstract_file": "chemprot_training_abstracts.tsv",
192
- "entity_file": "chemprot_training_entities.tsv",
193
- "relation_file": "chemprot_training_relations.tsv",
194
- "gold_standard_file": "chemprot_training_gold_standard.tsv",
195
- "split": "train",
196
- },
197
- ),
198
- datasets.SplitGenerator(
199
- name=datasets.Split.TEST,
200
- gen_kwargs={
201
- "filepath": os.path.join(test_path, "chemprot_test_gs"),
202
- "abstract_file": "chemprot_test_abstracts_gs.tsv",
203
- "entity_file": "chemprot_test_entities_gs.tsv",
204
- "relation_file": "chemprot_test_relations_gs.tsv",
205
- "gold_standard_file": "chemprot_test_gold_standard.tsv",
206
- "split": "test",
207
- },
208
- ),
209
- datasets.SplitGenerator(
210
- name=datasets.Split.VALIDATION,
211
- gen_kwargs={
212
- "filepath": os.path.join(dev_path, "chemprot_development"),
213
- "abstract_file": "chemprot_development_abstracts.tsv",
214
- "entity_file": "chemprot_development_entities.tsv",
215
- "relation_file": "chemprot_development_relations.tsv",
216
- "gold_standard_file": "chemprot_development_gold_standard.tsv",
217
- "split": "dev",
218
- },
219
- ),
220
- ]
221
-
222
- def _generate_examples(
223
- self,
224
- filepath,
225
- abstract_file,
226
- entity_file,
227
- relation_file,
228
- gold_standard_file,
229
- split,
230
- ):
231
- """Yields examples as (key, example) tuples."""
232
- if self.config.schema == "source":
233
- abstracts = self._get_abstract(os.path.join(filepath, abstract_file))
234
-
235
- entities, entity_id = self._get_entities(
236
- os.path.join(filepath, entity_file)
237
- )
238
-
239
- if self.config.subset_id == "chemprot_full":
240
- relations = self._get_relations(os.path.join(filepath, relation_file))
241
- elif self.config.subset_id == "chemprot_shared_task_eval":
242
- relations = self._get_relations_gs(
243
- os.path.join(filepath, gold_standard_file)
244
- )
245
- else:
246
- raise ValueError(self.config)
247
-
248
- for id_, pmid in enumerate(abstracts.keys()):
249
- yield id_, {
250
- "pmid": pmid,
251
- "text": abstracts[pmid],
252
- "entities": entities[pmid],
253
- "relations": relations.get(pmid, []),
254
- }
255
-
256
- elif self.config.schema == "bigbio_kb":
257
-
258
- abstracts = self._get_abstract(os.path.join(filepath, abstract_file))
259
- entities, entity_id = self._get_entities(
260
- os.path.join(filepath, entity_file)
261
- )
262
- relations = self._get_relations(
263
- os.path.join(filepath, relation_file), is_mapped=True
264
- )
265
-
266
- uid = 0
267
- for id_, pmid in enumerate(abstracts.keys()):
268
- data = {
269
- "id": str(uid),
270
- "document_id": str(pmid),
271
- "passages": [],
272
- "entities": [],
273
- "relations": [],
274
- "events": [],
275
- "coreferences": [],
276
- }
277
- uid += 1
278
-
279
- data["passages"] = [
280
- {
281
- "id": str(uid),
282
- "type": "title and abstract",
283
- "text": [abstracts[pmid]],
284
- "offsets": [[0, len(abstracts[pmid])]],
285
- }
286
- ]
287
- uid += 1
288
-
289
- entity_to_id = {}
290
- for entity in entities[pmid]:
291
- _text = entity["text"]
292
- entity.update({"text": [_text]})
293
- entity_to_id[entity["id"]] = str(uid)
294
- entity.update({"id": str(uid)})
295
- _offsets = entity["offsets"]
296
- entity.update({"offsets": [_offsets]})
297
- entity["normalized"] = []
298
- data["entities"].append(entity)
299
- uid += 1
300
-
301
- for relation in relations.get(pmid, []):
302
- relation["arg1_id"] = entity_to_id[relation.pop("arg1")]
303
- relation["arg2_id"] = entity_to_id[relation.pop("arg2")]
304
- relation.update({"id": str(uid)})
305
- relation["normalized"] = []
306
- data["relations"].append(relation)
307
- uid += 1
308
-
309
- yield id_, data
310
-
311
- @staticmethod
312
- def _get_abstract(abs_filename: str) -> Dict[str, str]:
313
- """
314
- For each document in PubMed ID (PMID) in the ChemProt abstract data file, return the abstract. Data is tab-separated.
315
-
316
- :param filename: `*_abstracts.tsv from ChemProt
317
-
318
- :returns Dictionary with PMID keys and abstract text as values.
319
- """
320
- with open(abs_filename, "r") as f:
321
- contents = [i.strip() for i in f.readlines()]
322
-
323
- # PMID is the first column, Abstract is last
324
- return {
325
- doc.split("\t")[0]: "\n".join(doc.split("\t")[1:]) for doc in contents
326
- } # Includes title as line 1
327
-
328
- @staticmethod
329
- def _get_entities(ents_filename: str) -> Tuple[Dict[str, str]]:
330
- """
331
- For each document in the corpus, return entity annotations per PMID.
332
- Each column in the entity file is as follows:
333
- (1) PMID
334
- (2) Entity Number
335
- (3) Entity Type (Chemical, Gene-Y, Gene-N)
336
- (4) Start index
337
- (5) End index
338
- (6) Actual text of entity
339
-
340
- :param ents_filename: `_*entities.tsv` file from ChemProt
341
-
342
- :returns: Dictionary with PMID keys and entity annotations.
343
- """
344
- with open(ents_filename, "r") as f:
345
- contents = [i.strip() for i in f.readlines()]
346
-
347
- entities = {}
348
- entity_id = {}
349
-
350
- for line in contents:
351
-
352
- pmid, idx, label, start_offset, end_offset, name = line.split("\t")
353
-
354
- # Populate entity dictionary
355
- if pmid not in entities:
356
- entities[pmid] = []
357
-
358
- ann = {
359
- "offsets": [int(start_offset), int(end_offset)],
360
- "text": name,
361
- "type": label,
362
- "id": idx,
363
- }
364
-
365
- entities[pmid].append(ann)
366
-
367
- # Populate entity mapping
368
- entity_id.update({idx: name})
369
-
370
- return entities, entity_id
371
-
372
- @staticmethod
373
- def _get_relations(rel_filename: str, is_mapped: bool = False) -> Dict[str, str]:
374
- """For each document in the ChemProt corpus, create an annotation for all relationships.
375
-
376
- :param is_mapped: Whether to convert into NL the relation tags. Default is OFF
377
- """
378
- with open(rel_filename, "r") as f:
379
- contents = [i.strip() for i in f.readlines()]
380
-
381
- relations = {}
382
-
383
- for line in contents:
384
- pmid, label, _, _, arg1, arg2 = line.split("\t")
385
- arg1 = arg1.split("Arg1:")[-1]
386
- arg2 = arg2.split("Arg2:")[-1]
387
-
388
- if pmid not in relations:
389
- relations[pmid] = []
390
-
391
- if is_mapped:
392
- label = _GROUP_LABELS[label]
393
-
394
- ann = {
395
- "type": label,
396
- "arg1": arg1,
397
- "arg2": arg2,
398
- }
399
-
400
- relations[pmid].append(ann)
401
-
402
- return relations
403
-
404
- @staticmethod
405
- def _get_relations_gs(rel_filename: str, is_mapped: bool = False) -> Dict[str, str]:
406
- """
407
- For each document in the ChemProt corpus, create an annotation for the gold-standard relationships.
408
-
409
- The columns include:
410
- (1) PMID
411
- (2) Relationship Label (CPR)
412
- (3) Used in shared task
413
- (3) Interactor Argument 1 Entity Identifier
414
- (4) Interactor Argument 2 Entity Identifier
415
-
416
- Gold standard includes CPRs 3-9. Relationships are always Gene + Protein.
417
- Unlike entities, there is no counter, hence once must be made
418
-
419
- :param rel_filename: Gold standard file name
420
- :param ent_dict: Entity Identifier to text
421
- """
422
- with open(rel_filename, "r") as f:
423
- contents = [i.strip() for i in f.readlines()]
424
-
425
- relations = {}
426
-
427
- for line in contents:
428
- pmid, label, arg1, arg2 = line.split("\t")
429
- arg1 = arg1.split("Arg1:")[-1]
430
- arg2 = arg2.split("Arg2:")[-1]
431
-
432
- if pmid not in relations:
433
- relations[pmid] = []
434
-
435
- if is_mapped:
436
- label = _GROUP_LABELS[label]
437
-
438
- ann = {
439
- "type": label,
440
- "arg1": arg1,
441
- "arg2": arg2,
442
- }
443
-
444
- relations[pmid].append(ann)
445
-
446
- return relations