-
Notifications
You must be signed in to change notification settings - Fork 0
/
Validation.py
405 lines (332 loc) · 18.6 KB
/
Validation.py
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
'''
Validation class
1. K-fold Cross Validation
Author: Kyoung Tak Cho
Created: Tue Sep 4 04:03:20 CDT 2018
Updated: Mon Jun 10 17:08:10 CDT 2019
'''
import settings
import sqls
import math
from models import ValidationDataset, FreqDict, GenePrediction, ConfusionMatrix
from GeneGroups import FreqDictSet, Genes
from utils.Common import ListTool
from utils.DBManager import Pgsql
#
# Validation - Kmer Frequency group for validation
#
class ValidationFreqDict(object):
def __init__(self, datasets=None, k=None, genes=None):
if datasets is None:
error_mesg = 'datasets is empty.'
raise ValueError(error_mesg)
if k is None:
error_mesg = 'k (kmer window size) is empty.'
raise ValueError(error_mesg)
self.datasets = datasets
self.class_size = len(self.datasets)
self.k = k
if genes is None:
self.genes = Genes()
else:
self.genes = genes
self.freq_dict_set = list() # FreqDictSet * class_size
#
# Initialization
#
# Build kmer frequency dictionary for each fold
self._init_freq_dict_set()
# Build kmer frequency dictionary for each fold
def _init_freq_dict_set(self):
for class_num, v_dataset in enumerate(self.datasets):
print('class_num: ', class_num)
self.freq_dict_set.append(list())
self.freq_dict_set[class_num] = FreqDictSet()
for fold_idx, fold in enumerate(v_dataset):
print('fold_idx: ', fold_idx, 'len: ', len(fold))
fd_k = FreqDict(k=self.k)
fd_k_1 = FreqDict(k=self.k - 1)
for gnid in fold:
gene_info = self.genes.get_gene(gnid=gnid)
# for k-mer
gene_fd_k = gene_info.pep_seq_max.get_kmer_freq(self.k)
for kmer, freq in gene_fd_k.kmer_freq.items():
fd_k.add_kmer_freq(kmer, freq)
# for (k-1)-mer
gene_fd_k_1 = gene_info.pep_seq_max.get_kmer_freq(self.k - 1)
for kmer, freq in gene_fd_k_1.kmer_freq.items():
fd_k_1.add_kmer_freq(kmer, freq)
self.freq_dict_set[class_num].append_kmer_dict(k=self.k, freq_dict=fd_k)
self.freq_dict_set[class_num].append_kmer_dict(k=self.k - 1, freq_dict=fd_k_1)
def get_genes(self):
return self.genes
def get_fd_class(self, class_num=None, k=None):
if class_num is None:
error_mesg = 'class num is empty.'
raise ValueError(error_mesg)
if k is None:
error_mesg = 'k is empty'
raise ValueError(error_mesg)
return self.freq_dict_set[class_num].get_fd_class(k=k)
def get_fd_fold(self, class_num=None, k=None, fold_idx=None):
if class_num is None:
error_mesg = 'class num is empty.'
raise ValueError(error_mesg)
if k is None:
error_mesg = 'k is empty'
raise ValueError(error_mesg)
if fold_idx is None:
error_mesg = 'fold_idx is empty'
raise ValueError(error_mesg)
return self.freq_dict_set[class_num].get_fd_fold(k=k, fold_idx=fold_idx)
def get_train_set(self, class_num=None, k=None, fold_idx=None):
return self.freq_dict_set[class_num].get_fd_complement(k=k, fold_idx=fold_idx)
def test_freq_dict_set(self):
for class_num, v_dataset in enumerate(self.datasets):
print('TEST - class_num: ', class_num)
self.freq_dict_set.append(list())
self.freq_dict_set[class_num] = FreqDictSet()
for fold_idx, fold in enumerate(v_dataset):
print('TEST - fold_idx: ', fold_idx, 'len: ', len(fold),)
fd_k = FreqDict(k=self.k)
fd_k_1 = FreqDict(k=self.k - 1)
for gnid in fold:
gene_info = self.genes.get_gene(gnid=gnid)
# for k-mer
gene_fd_k = gene_info.pep_seq_max.get_kmer_freq(self.k)
for kmer, freq in gene_fd_k.kmer_freq.items():
fd_k.add_kmer_freq(kmer, freq)
# for (k-1)-mer
gene_fd_k_1 = gene_info.pep_seq_max.get_kmer_freq(self.k - 1)
for kmer, freq in gene_fd_k_1.kmer_freq.items():
fd_k_1.add_kmer_freq(kmer, freq)
print('fd_k value sum: {}'.format(fd_k.total_freq()))
print('fd_k_1 value sum: {}'.format(fd_k_1.total_freq()))
self.freq_dict_set[class_num].append_kmer_dict(k=self.k, freq_dict=fd_k)
self.freq_dict_set[class_num].append_kmer_dict(k=self.k - 1, freq_dict=fd_k_1)
#
# K-fold Cross Validation
#
class CrossValidation(object):
def __init__(self, genes=None, all_gnids=None,
class_size=None, fold_size=None, kmer_size=None,
exp_setting=None):
if exp_setting is not None and not exp_setting.get_test_mode():
if genes is None:
error_mesg = 'genes is empty.'
raise ValueError(error_mesg)
if class_size is None:
error_mesg = 'class size is empty.'
raise ValueError(error_mesg)
if fold_size is None:
error_mesg = 'fold size is empty.'
raise ValueError(error_mesg)
if kmer_size is None:
error_mesg = 'kmer size is empty.'
raise ValueError(error_mesg)
# general attributes
self.genes = genes # Genes type
self.all_gnids = all_gnids # list() type
self.datasets = list() # ValidationDataset
self.class_size = class_size
self.fold_size = fold_size
self.kmer_size = kmer_size
self.prediction_results = dict() # * number of gnids
self.assigned_genes = list()
self.exp_setting = exp_setting
def build_datasets(self, assigned_genes=None, neg_class_mode=1, corresp_tissue=None):
if assigned_genes is None:
error_mesg = 'assigned is empty.'
raise ValueError(error_mesg)
else: # build assigned gnids for promoter data
if self.exp_setting.get_seq_type() == 'm1':
# remove missing gnids for promoter data
self.assigned_genes = self.remove_missing_gnids(assigned_genes)
# set exclusive gnids
exclusive_gnids = self.get_exclusive_gnids(assigned_genes)
else:
self.assigned_genes = assigned_genes.copy()
exclusive_gnids = self.assigned_genes
if corresp_tissue is None:
raise ValueError('corresp_tissue is empty.')
if self.fold_size is None:
self.fold_size = 1
print('fold size is empty. fold_size = 1 (default)')
if self.fold_size <= 0:
self.fold_size = 1
print('fold size is less than zero. fold_size = 1 (default)')
for class_num in range(0, self.class_size):
if settings.NEED_NEW_GENE_DATA:
self.datasets.append(class_num)
self.datasets[class_num] = list()
sub_dataset = list()
gp_type = self.exp_setting.get_gp_type()
if class_num == 0:
if neg_class_mode == settings.NEG_CLASS_MODE_NOT_P:
wd_all_gnids_per_tissue = GetData.wd_all_gnid_per_tissue(self.exp_setting, corresp_tissue)
sub_dataset = list(set(wd_all_gnids_per_tissue) - set(self.assigned_genes))
elif neg_class_mode == settings.NEG_CLASS_MODE_RND_S: # random and same number with POS class
if gp_type not in ('g', 'p'): # default is type is 'p' for gp_type == 'm1' (combined type)
gp_type = 'p'
pars = (gp_type, corresp_tissue, ListTool.list2str(exclusive_gnids, ','), len(self.assigned_genes))
random_assigned_gene = Pgsql.Common.select_data(sqls.get_gene_tissues_random, pars)
sub_dataset = ListTool.twoD2oneD(random_assigned_gene)
elif neg_class_mode == settings.NEG_CLASS_MODE_RND_M:
wd_all_gnids_per_tissue = GetData.wd_all_gnid_per_tissue(self.exp_setting, corresp_tissue)
sub_dataset = list(set(wd_all_gnids_per_tissue) - set(self.assigned_genes))
else: # default
wd_all_gnids_per_tissue = GetData.wd_all_gnid_per_tissue(self.exp_setting, corresp_tissue)
sub_dataset = list(set(wd_all_gnids_per_tissue) - set(self.assigned_genes))
# set negative class genes
self.exp_setting.set_gene_dataset(feature_id=corresp_tissue, negative_class=sub_dataset)
elif class_num == 1:
sub_dataset = self.assigned_genes
# set positive class genes
self.exp_setting.set_gene_dataset(feature_id=corresp_tissue, positive_class=sub_dataset)
self.datasets[class_num] = ValidationDataset(gnids=sub_dataset, fold_size=self.fold_size)
else:
sub_dataset_neg = self.exp_setting.get_gene_dataset_neg()
sub_dataset_pos = self.exp_setting.get_gene_dataset_pos()
self.datasets[0] = ValidationDataset(gnids=sub_dataset_neg, fold_size=self.fold_size)
self.datasets[1] = ValidationDataset(gnids=sub_dataset_pos, fold_size=self.fold_size)
def remove_missing_gnids(self, all_gnids):
# remove missing gnids for promoter
missing_gnids = self.exp_setting.get_missing_gnids_in_promoter()
#new_sub_dataset = list(set(all_gnids) - set(missing_gnids))
new_sub_dataset = ListTool.sub(all_gnids, missing_gnids)
print('all gnids:', len(all_gnids))
print('common items:', len(ListTool.common_items(all_gnids, missing_gnids)))
print('removed:', len(new_sub_dataset))
return sorted(new_sub_dataset)
def get_exclusive_gnids(self, all_gnids):
missing_gnids = self.exp_setting.get_missing_gnids_in_promoter()
exclusive_gnids = ListTool.add_rm_dup(all_gnids, missing_gnids)
return sorted(exclusive_gnids)
def test_datasets(self):
if len(self.datasets) == 0:
raise ValueError('dataset is empty.')
for class_num, dataset in enumerate(self.datasets):
for fold_idx, fold in enumerate(dataset):
print('fold#: {}, genes: {}'.format(fold_idx, ','.join(str(gnid) for gnid in fold)))
def validation(self):
#
# Phase 1: build Kmer Frequency dictionary for each fold
#
print('Step 1 - build frequency dictionaries')
v_freq_dict = ValidationFreqDict(datasets=self.datasets, k=self.kmer_size, genes=self.genes)
# update genes info
self.genes = v_freq_dict.get_genes()
# TEST
if self.exp_setting.get_test_mode() == settings.TEST_MODE_KMER_FREQ:
v_freq_dict.test_freq_dict_set()
#
# Phase 2: compute probabilities for each fold
#
print('Stepe 2 - prediction')
# Gene prediction results dictionary
prediction_results = dict()
# build frequency dictionaries by class level
fd_class_k = list()
fd_class_k_1 = list()
for class_num in range(0, self.class_size):
fd_class_k.append(v_freq_dict.get_fd_class(class_num=class_num, k=self.kmer_size))
fd_class_k_1.append(v_freq_dict.get_fd_class(class_num=class_num, k=self.kmer_size - 1))
print('total: fd_class_k:', fd_class_k[class_num].total_freq_float(), len(fd_class_k[class_num]), len(fd_class_k))
print('total: fd_class_k-1:', fd_class_k_1[class_num].total_freq_float(), len(fd_class_k_1[class_num]), len(fd_class_k_1))
# set dataset size for each class and compute probability for each class
p_class = [0., 0.]
size_total = len(self.datasets[0]) + len(self.datasets[1])
for class_num in range(0, self.class_size):
p_class[class_num] = len(self.datasets[class_num]) / size_total
for assigned_class, dataset in enumerate(self.datasets):
for fold_idx, fold in enumerate(dataset):
if self.exp_setting.get_test_mode() == settings.TEST_MODE_KMER_FREQ:
print('Total gnids in fold# {}: {}, gnids: {}'.format(fold_idx,
len(fold),
','.join(str(gnid) for gnid in fold)))
fd_train_set_k = list() # * class_size
fd_train_set_k_1 = list() # * class_size
fd_train_set_total_freq_k = list()
fd_train_set_total_freq_k_1 = list()
for class_num in range(0, self.class_size):
fd_train_set_k.append(fd_class_k[class_num] - v_freq_dict.get_fd_fold(class_num=class_num, k=self.kmer_size, fold_idx=fold_idx))
fd_train_set_k_1.append(fd_class_k_1[class_num] - v_freq_dict.get_fd_fold(class_num=class_num, k=self.kmer_size - 1, fold_idx=fold_idx))
fd_train_set_total_freq_k.append(fd_train_set_k[class_num].total_freq_float())
fd_train_set_total_freq_k_1.append(fd_train_set_k_1[class_num].total_freq_float())
# initialize confusion matrix
cm = ConfusionMatrix()
# check for both classes
if fd_train_set_total_freq_k[class_num] == 0 or fd_train_set_total_freq_k_1[class_num] == 0:
dataset.set_confusion_matrix(cm=cm, fold_idx=fold_idx)
prediction_results = dict()
return prediction_results
for gnid in fold:
gene_info = self.genes.get_gene(gnid=gnid)
fd_gene_k = gene_info.pep_seq_max.get_kmer_freq(self.kmer_size)
fd_gene_k_1 = gene_info.pep_seq_max.get_kmer_freq(self.kmer_size - 1)
log_p = [0., 0.]
# 3-mer
for kmer in fd_gene_k.kmer_freq:
freq_k = fd_gene_k.kmer_freq[kmer]
if freq_k > 0:
for class_num in range(0, self.class_size):
freq_k_in_train = fd_train_set_k[class_num].get_kmer_freq_value(kmer=kmer)
freq_k_total = fd_train_set_total_freq_k[class_num]
prob = freq_k * math.log(freq_k_in_train / freq_k_total)
log_p[class_num] += prob
# TEST - KmerFreq/FreqDict
if self.exp_setting.get_test_mode() == settings.TEST_MODE_KMER_FREQ:
print('class#: {}, kmer: {}, freq_k: {}, probability: {}, freq_k in train set: {}, freq_k total: {}'.format(
class_num, kmer, freq_k, prob, freq_k_in_train, freq_k_total))
else:
print('kmer: {}, frequency: {}'.format(kmer, freq_k))
raise(ValueError('kmer frequency is less than 0'))
# 2-mer
for kmer in fd_gene_k_1.kmer_freq:
freq_k_1 = fd_gene_k_1.kmer_freq[kmer]
if freq_k_1 > 0:
for class_num in range(0, self.class_size):
freq_k_1_in_train = fd_train_set_k_1[class_num].get_kmer_freq_value(kmer=kmer)
freq_k_1_total = fd_train_set_total_freq_k_1[class_num]
prob = freq_k_1 * math.log(freq_k_1_in_train / freq_k_1_total)
log_p[class_num] -= prob
# TEST - KmerFreq/FreqDict
if self.exp_setting.get_test_mode() == settings.TEST_MODE_KMER_FREQ:
print('class#: {}, kmer: {}, freq_k: {}, probability: {}, freq_k_1 in train set: {}, freq_k_1 total: {}'.format(
class_num, kmer, freq_k_1, prob, freq_k_1_in_train, freq_k_1_total))
# log_p finalize - adding probability for each class
for class_num in range(0, self.class_size):
log_p[class_num] += math.log(p_class[class_num])
# TEST - KmerFreq/FreqDict
if self.exp_setting.get_test_mode() == settings.TEST_MODE_KMER_FREQ:
print('seq: {}'.format(gene_info.pep_seq_max.get_seq_str()))
print('log_p[0]: {}, log_p[1]: {}'. format(log_p[0], log_p[1]))
# set prediction results
gene_prediction = prediction_results.get(gnid, None)
if gene_prediction is None:
gene_prediction = GenePrediction(gnid=gnid, class_size=self.class_size, assigned_class=assigned_class)
if log_p[0] > log_p[1]:
predicted_class = 0
# set confusion_matrix
if assigned_class == 0:
cm.add_tn()
else:
cm.add_fn()
else:
predicted_class = 1
# set confusion_matrix
if assigned_class == 0:
cm.add_fp()
else:
cm.add_tp()
# set predicted class
gene_prediction.set_predicted_class(predicted_class=predicted_class, log_p=log_p)
prediction_results[gnid] = gene_prediction
# set confusion matrix for each fold
dataset.set_confusion_matrix(cm=cm, fold_idx=fold_idx)
# TEST
print ('prediction results length: ', len(prediction_results))
#return sorted(prediction_results)
return prediction_results
def get_fold_size(self):
return self.fold_size()