Repository navigation
Expand file tree
/
Copy pathpreprocess.py
More file actions
493 lines (444 loc) · 17.9 KB
/
Copy pathpreprocess.py
File metadata and controls
493 lines (444 loc) · 17.9 KB
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
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
from __future__ import division
import pickle as pkl
import obonet
import json
import numpy as np
import re
import string
import random
from gensim import models, corpora, matutils
from nltk.tokenize import word_tokenize
from nltk.corpus import stopwords
from nltk.stem.porter import PorterStemmer
from scipy.sparse import csc_matrix, hstack, vstack
from Bio.Entrez import efetch
data_directory = "../data/"
labels = ["DB","DB_OID","DB_OBS",
"Qualifier","GO_ID","DB_REF","EVIDENCE",
"WITH","Aspect","DB_OBN","DB_OBSYN",
"DB_OBType","Taxon","Date", "Assigned_By",
"Annotation_EXT","Gene_PFID"]
evidence_codes = ["EXP", "IDA", "IPI",
"IMP", "IGI", "IEP",
"HTP", "HDA", "HMP",
"HGI", "HEP",
"ISS", "ISO", "ISA",
"ISM", "IGC", "IBA",
"IBD", "IKR", "IRD",
"RCA", "TAS", "NAS",
"IC", "ND"]
feature_labels = {"GO_ID", "DB_REF", "GO_PARENTS", "Aspect", "Taxon", "DB_OBSYN", "WITH"}
stoplist = set('for a of the and to in'.split())
regex = re.compile('[%s]' % re.escape(string.punctuation))
port_stemmer = PorterStemmer()
def load_pkl_data(filename):
f = open(filename, 'rb')
data = pkl.load(f)
return data
def save_json_data(filename, data, indent=None):
with open(filename, 'w') as f:
json.dump(data, f, indent=indent)
def save_pkl_data(filename, data):
with open(filename, 'wb') as f:
pkl.dump(data, f)
def load_json_data(filename):
f = open(filename)
data = json.load(f)
return data
def dump_to_json(filename, outfile=None):
# dat = load_data("../data/new_annotations.pkl")
dat = load_pkl_data(filename)
data_dict = []
for each in dat:
curr = {}
for val, label in zip(each, labels):
curr[label] = val
data_dict.append(curr)
if outfile is None:
outfile = data_directory+"all_data.json"
with open(outfile,'w') as f:
json.dump(data_dict, f)
def load_pubmed_text_data(filename=None):
if not filename:
filename = data_directory+"pubmed_output_1.txt"
data = {}
with open(filename, 'r', encoding='latin1')as f:
for each in f:
splitted = each.split('\t',2)
pubid = splitted[0]
pubyr = splitted[1]
abstract = splitted[2]
data[pubid] = {"date": pubyr, "abstract": abstract}
return data
def dump_pubmed_json_fromtext(infile=None, outfile=None):
data = load_pubmed_text_data(infile)
if not outfile:
outfile = data_directory + "pubmed.json"
with open(outfile, 'w') as f:
json.dump(data,f, indent=2)
def create_go_term_vector(filename=None, dump=False):
if not filename:
filename = data_directory+ "go.obo"
graph = obonet.read_obo(filename)
nodes = graph._adj.keys()
node_to_index = {}
index_to_node = {}
for i, node in enumerate(nodes):
node_to_index[node] = i
index_to_node[i] = node
# ohenc = OneHotEncoder(n_values=len(node_to_index))
# data = ohenc.fit([[1],[3456],[234],[4367]])
if dump:
with open(data_directory + "node_to_index.pkl", 'wb') as f:
pkl.dump(node_to_index, f)
with open(data_directory + "index_to_node.pkl", 'wb') as f:
pkl.dump(index_to_node, f)
return node_to_index, index_to_node
def get_parent_nodes(node, graph=None):
if not graph:
graph = obonet.read_obo(data_directory + "go.obo")
ans = []
node_dic = graph._adj[node]
for every in node_dic:
for every_key in node_dic[every]:
if every_key == "is_a":
ans.append(every)
return ans
def collect_cleaned_goref_pubmed_data(pumed_json_filename, go_ref=False, dump=False):
all_text = []
all_text_dict = {}
if go_ref:
with open(data_directory + "GO_REF.pickle", 'rb') as f:
gorefData = pkl.load(f)
for i, each in enumerate(gorefData):
text = gorefData[each]
for line in text:
line = clean_text(line)
all_text.append(line)
all_text_dict[each] = line
pubmed_data = load_json_data(pumed_json_filename)
for each in pubmed_data:
line = clean_text(pubmed_data[each]["abstract"])
all_text.append(line)
all_text_dict[each] = line
if dump:
with open(data_directory + "all_abstract.pickle",'wb') as f:
pkl.dump(all_text, f)
with open(data_directory + "all_abstract_withID.pickle", 'wb') as f:
pkl.dump(all_text_dict, f)
return all_text, all_text_dict
def clean_text(line, stemming=False):
line = line.lower()
# remove urls (courtesy : stackoverflow)
line = re.sub(
r'(?i)\b((?:https?://|www\d{0,3}[.]|[a-z0-9.\-]+[.][a-z]{2,4} /)(?:[^\s()<>]+|\(([^\s()<>]+|(\([^\s()<>]+\)))*\))+(?:\(([^\s()<>]+|(\([^\s()<>]+\)))*\)|[^\s`!()\[\]{};:\'".,<>?«»“”‘’]))',
'', line)
new_line = []
# tokenizing
tokenized_line = word_tokenize(text=line)
for token in tokenized_line:
# removing punctuation and stop words
if '-' in token and len(token) > 3:
token = token.split('-')
new_token = ' '.join(token)
else:
new_token = regex.sub(u'', token)
if not new_token == u'' and not new_token in stopwords.words('english') and new_token not in stoplist:
if stemming:
new_token = port_stemmer.stem(new_token)
new_line.append(new_token)
return ' '.join(new_line)
def get_evidence_code_dict():
evid = {}
for i, each in enumerate(evidence_codes):
evid[each] = i
return evid
def get_corpus(texts, dct=None):
# use a pre-saved dictionary for abstracts
if not dct:
dct = load_pkl_data(data_directory + "dct.pickle")
corpus = [dct.doc2bow(line) for line in texts]
return corpus
def get_tfidf_vectors_sparse(corpus, tfidf_model=None):
if not tfidf_model:
tfidf_model = load_pkl_data(data_directory + "tfidf_model.pickle")
vectors = []
num_terms = len(tfidf_model.idfs)
for i , each in enumerate(corpus):
vector = tfidf_model[corpus[i]]
vectors.append(vector)
scipy_csc_matrix = matutils.corpus2csc(vectors, num_terms=num_terms)
return scipy_csc_matrix.T
def create_tfidf_model(documents=None, dump=False):
if not documents:
documents = load_pkl_data(data_directory + "all_abstract.pickle")
texts = [[word for word in document.lower().split() if word not in stoplist]
for document in documents]
dct = corpora.Dictionary(texts)
corpus = [dct.doc2bow(line) for line in texts] # convert dataset to BoW format
model = models.TfidfModel(corpus) # fit model
if dump:
save_pkl_data(data_directory + "dct.pickle", dct)
save_pkl_data(data_directory + "tfidf_model.pickle", model)
return model
def create_training_data(data_filename, ontology_filename=None, dump=False, feature_filename=None, label_filename=None):
node_to_index, index_to_node = create_go_term_vector(ontology_filename)
json_data = load_json_data(data_filename)
graph = obonet.read_obo(data_directory + "go.obo")
abstract_data = load_pkl_data(data_directory + "all_abstract_withID.pickle")
dct = load_pkl_data(data_directory + "dct.pickle")
tfidf_model = load_pkl_data(data_directory + "tfidf_model.pickle")
evid_dict = get_evidence_code_dict()
lab = {"GO_ID", "DB_REF", "GO_PARENTS", "Aspect", "Taxon", "DB_OBSYN", "WITH"}
Aspect = {"P":0, "F":1, "C":2}
features = []
labels = []
for point in json_data:
#GO VECTOR
goterm = point["GO_ID"]
go_one_hot = np.zeros(len(node_to_index))
go_one_hot[node_to_index[goterm]] = 1.0
go_one_hot = csc_matrix(go_one_hot)
# ABSTRACT VECTOR
dbref = point["DB_REF"]
if "PMID" in dbref:
dbref = dbref.split(':')
dbref = dbref[1]
abstract = abstract_data[dbref]
abstract = [[word for word in abstract.lower().split() if word not in stoplist]]
corpus = get_corpus(abstract, dct)
abstract_vec = get_tfidf_vectors_sparse(corpus, tfidf_model)
feature = hstack([go_one_hot, abstract_vec])
#GO_PARENTS
parents = get_parent_nodes(goterm, graph)
parent_vec = np.zeros(len(node_to_index))
for each in parents:
parid =node_to_index[each]
parent_vec[parid] = 1.0
parent_vec = csc_matrix(parent_vec)
feature = hstack([feature, parent_vec])
#ASPECT:
aspect_one_hot = np.zeros(len(Aspect))
aspect_one_hot[Aspect[point["Aspect"]]] = 1.0
aspect_one_hot = csc_matrix(aspect_one_hot)
feature = hstack([feature, aspect_one_hot])
#Label
evd_c = point["EVIDENCE"]
# evd_one_hot = np.zeros(len(evidence_codes))
# evd_one_hot[evid_dict[evd_c]] = 1.0
# evd_one_hot = csc_matrix(evd_one_hot)
#stack all data
features.append(feature)
labels.append(evid_dict[evd_c])
features = vstack(features)
labels = np.asarray(labels)
if dump:
if feature_filename is None:
feature_filename = "data_feature_vector.pickle"
save_pkl_data(data_directory + feature_filename, features)
if label_filename is None:
label_filename = "data_labels_vector_non_sparse.pickle"
save_pkl_data(data_directory + label_filename, labels)
def get_abstracts_from_pmid(pmid):
handle = efetch(db='pubmed', id=pmid, retmode='text', rettype='abstract')
return (handle.read())
def create_raw_data(filename=None):
data_unique_annotations_pmids = []
data_complex_pmids = []
if filename is None:
filename = "../data/all_data.json"
all_data = load_json_data(filename)
available_pmid = load_json_data("../data/pubmed.json")
u_pmid = load_pkl_data("../data/unique_PMID.pickle")
pmid_not_present = set()
is_presence_of_extra_pmids = False
for datapoint in all_data:
if "PMID" in datapoint["DB_REF"]:
pmid = datapoint["DB_REF"]
pmid = pmid.split(':')
pmid = pmid[1]
if pmid in available_pmid:
data_complex_pmids.append(datapoint)
if datapoint["DB_REF"] in u_pmid:
data_unique_annotations_pmids.append(datapoint)
else:
try:
is_presence_of_extra_pmids = True
abstract = get_abstracts_from_pmid(pmid)
if abstract is not None and len(abstract) > 0:
available_pmid[pmid] = {}
available_pmid[pmid]["abstract"] = abstract
data_complex_pmids.append(datapoint)
if datapoint["DB_REF"] in u_pmid:
data_unique_annotations_pmids.append(datapoint)
else:
print("IN TRY!! COULD NOT FIND PMID GLOBALLY")
pmid_not_present.add(pmid)
except:
print("COULD NOT FIND PMID GLOBALLY")
pmid_not_present.add(pmid)
save_json_data("../data/pmids_not_present.json", list(pmid_not_present))
save_json_data("../data/pubmed_latest.json", available_pmid, indent=1)
print("NUMBER OF PMIDS NOT AVAILABLE ARE: %d"%len(pmid_not_present))
print("TOTAL DATA POINTS FOR COMPLEX DATA IS %d"%len(data_complex_pmids))
print("TOTAL DATA POINTS FOR UNIQUE ANNOTATION PMID IS %d"%len(data_unique_annotations_pmids))
save_json_data("../data/data_with_all_pmids.json", data_complex_pmids)
save_json_data("../data/data_with_uniqueAnnotation_pmid.json", data_unique_annotations_pmids)
def create_unique_abstract_dict():
unique = set()
d = load_pkl_data("../data/dbref_annotations.pickle")
count = 0
for each in d:
if "PMID" in each:
if d[each] == 1:
count += 1
unique.add(each)
save_pkl_data("../data/unique_PMID.pickle", unique)
print("Number of PMIDS with unique annotation are: %d"%count)
def create_train_dev_test():
unique_data = load_json_data("../data/data_with_uniqueAnnotation_pmid.json")
all_data = load_json_data("../data/data_with_all_pmids.json")
random.shuffle(unique_data)
random.shuffle(all_data)
udata_len = len(unique_data)
all_data_len = len(all_data)
#create train/dev/test for unique ones
Xtrain_unique, Xdev_unique, Xtest_unique = split_data_helper(udata_len, unique_data)
#create train/dev/test for all
Xtrain_all, Xdev_all, Xtest_all = split_data_helper(all_data_len, all_data)
save_json_data("../data/final_data/Xtrain_unique.json", Xtrain_unique)
save_json_data("../data/final_data/Xdev_unique.json", Xdev_unique)
save_json_data("../data/final_data/Xtest_unique.json", Xtest_unique)
save_json_data("../data/final_data/Xtrain_all.json", Xtrain_all)
save_json_data("../data/final_data/Xdev_all.json", Xdev_all)
save_json_data("../data/final_data/Xtest_all.json", Xtest_all)
def split_data_helper(data_len, data):
train_len = int(0.7 * data_len)
rem_len = data_len - train_len
dev_len = int(rem_len * 2 / 3)
Xtrain = data[: train_len]
Xdev = data[train_len: train_len+dev_len]
Xtest = data[train_len+dev_len : ]
return Xtrain, Xdev, Xtest
# collect_cleaned_goref_pubmed_data(data_directory+"final_data/train_dev_test/pubmed_latest.json", dump=True)
# create_tfidf_model(dump=True)
# create_training_data("../data/final_data/train_dev_test/unique/Xdev_unique.json",dump=True,feature_filename="Xdev_unique.pickle",label_filename="Ydev_unique.pickle")
# create_training_data("../data/final_data/train_dev_test/all/Xdev_all.json",dump=True,feature_filename="Xdev.pickle",label_filename="Ydev.pickle")
# create_train_dev_test()
# create_raw_data()
# create_unique_abstract_dict()
# dump_pubmed_json_fromtext("../data/pubmed_output_1.txt")
# create_tfidf_model()
# collect_cleaned_goref_pubmed_data(dump=True)
# print("OK")
# documents = load_pkl_data("../data/all_abstract.pickle")
# texts = [[word for word in document.lower().split() if word not in stoplist]
# for document in documents]
# corpus = get_corpus(texts)
# vecs = get_tfidf_vectors_sparse(corpus)
# create_go_term_vector()
# create_training_data("../data/temp_data.json", dump=True)
# data = load_json_data(data_directory+"temp_data.json")
# print (len(data))
# json_data = load_json_data("../data/all_data.json")
# abstract_data = load_pkl_data("../data/all_abstract_withID_1.pickle")
# dbref_dict_evidence_code = {}
# dbref_annotations = {}
# gene_id_annotations = {}
# gene_id_evidence = {}
#
# count = 0
# pcount = 0
# pmid_count = 0
# go_ref_count = 0
# other_count = 0
# temp_data = []
# pmid_set_np = set()
# pmid_set_p = set()
#
# total_count = 0
# for point in json_data:
# total_count+=1
# dbref = point["DB_REF"]
# evidence = point["EVIDENCE"]
# geneID = point["GO_ID"]
# #####################
# if dbref in dbref_dict_evidence_code:
# dbref_dict_evidence_code[dbref].add(evidence)
# else:
# dbref_dict_evidence_code[dbref] = set(evidence)
# if dbref in dbref_annotations:
# dbref_annotations[dbref]+=1
# else:
# dbref_annotations[dbref] = 1
#
# if geneID in gene_id_evidence:
# gene_id_evidence[geneID].add(evidence)
# else:
# gene_id_evidence[geneID] = set(evidence)
# if geneID in gene_id_annotations:
# gene_id_annotations[geneID]+=1
# else:
# gene_id_annotations[geneID] = 1
#
# #####################
# if "PMID" in dbref:
# dbref = dbref.split(':')
# dbref = dbref[1]
# pmid_count+=1
# try:
# abstract = abstract_data[dbref]
# pmid_set_p.add(dbref)
# temp_data.append(point)
# except:
# pmid_set_np.add(dbref)
# elif "GO_REF" in dbref:
# go_ref_count+=1
# else:
# other_count+=1
#
# print("PMID COUNT IS %d which is %f of total count"%(pmid_count, pmid_count/total_count))
# print("GO_REF COUNT IS %d which is %f of total count"%(go_ref_count, go_ref_count/total_count))
# print("OTHERS COUNT IS %d which is %f of total count"%(other_count, other_count/total_count))
# print("TOTAL COUNT IS %d"%total_count)
#
# count_dbref_annotations = 0
# count_geneID_annotations = 0
# count_dbref_evidence_code = 0
# count_geneID_evidence_code = 0
#
# for each in dbref_annotations:
# count_dbref_annotations += dbref_annotations[each]
#
# for each in dbref_dict_evidence_code:
# count_dbref_evidence_code += len(dbref_dict_evidence_code[each])
#
# for each in gene_id_annotations:
# count_geneID_annotations += gene_id_annotations[each]
#
# for each in gene_id_evidence:
# count_geneID_evidence_code += len(gene_id_evidence[each])
#
# print("Average number of annotations per GENE ID is %f"%(count_geneID_annotations/len(gene_id_annotations)))
# print("Average number of evidence codes per GENE ID is %f"%(count_geneID_evidence_code/len(gene_id_evidence)))
#
# print("Average number of annotations per ABSTRACT ID is %f"%(count_dbref_annotations/len(dbref_annotations)))
# print("Average number of evidence codes per ABSTRACT ID is %f"%(count_dbref_evidence_code/len(dbref_dict_evidence_code)))
#
# print("MAX NUMBER OF ANNOTATIONS FOR AN ABSTRACT IS %d"%(max(dbref_annotations.values())))
# print("MAX NUMBER OF ANNOTATIONS FOR A GENE ID IS %d"%(max(gene_id_annotations.values())))
# save_pkl_data("../data/dbref_annotations.pickle", dbref_annotations)
# save_pkl_data("../data/gene_id_annotations.pickle", gene_id_annotations)
# d = load_pkl_data("../data/dbref_annotations.pickle")
# print("OK")
# print("NOT PRESENT ARE %d"%len(pmid_set_np))
# print("PRESENT ARE %d"%len(pmid_set_p))
# with open("not_present.txt", 'w') as f:
# for each in pmid_set_np:
# f.write(str(each)+"\n")
# save_json_data("../data/temp_data.json", temp_data)
# json_data = load_json_data("../data/temp_data.json")
# i = 0
# for each in json_data:
# i+=1
# print(i)