|
0
|
1 from __future__ import division
|
|
|
2 import os
|
|
|
3 import sys
|
|
|
4 import pandas as pd
|
|
|
5 import collections
|
|
|
6 import pickle as pk
|
|
|
7 import argparse
|
|
|
8 from sklearn.cluster import KMeans
|
|
13
|
9 import matplotlib
|
|
|
10 matplotlib.use('GTKAgg')
|
|
0
|
11 import matplotlib.pyplot as plt
|
|
|
12
|
|
|
13 ########################## argparse ###########################################
|
|
|
14
|
|
|
15 def process_args(args):
|
|
|
16 parser = argparse.ArgumentParser(usage = '%(prog)s [options]',
|
|
|
17 description = 'process some value\'s' +
|
|
|
18 ' genes to create class.')
|
|
|
19 parser.add_argument('-rs', '--rules_selector',
|
|
|
20 type = str,
|
|
|
21 default = 'HMRcore',
|
|
|
22 choices = ['HMRcore', 'Recon', 'Custom'],
|
|
|
23 help = 'chose which type of dataset you want use')
|
|
|
24 parser.add_argument('-cr', '--custom',
|
|
|
25 type = str,
|
|
|
26 help='your dataset if you want custom rules')
|
|
|
27 parser.add_argument('-ch', '--cond_hier',
|
|
|
28 type = str,
|
|
|
29 default = 'no',
|
|
|
30 choices = ['no', 'yes'],
|
|
|
31 help = 'chose if you wanna hierical dendrogram')
|
|
|
32 parser.add_argument('-lk', '--k_min',
|
|
|
33 type = int,
|
|
|
34 help = 'min number of cluster')
|
|
|
35 parser.add_argument('-uk', '--k_max',
|
|
|
36 type = int,
|
|
|
37 help = 'max number of cluster')
|
|
|
38 parser.add_argument('-li', '--linkage',
|
|
|
39 type = str,
|
|
|
40 choices = ['single', 'complete', 'average'],
|
|
|
41 help='linkage hierarchical cluster')
|
|
|
42 parser.add_argument('-d', '--data',
|
|
|
43 type = str,
|
|
|
44 required = True,
|
|
|
45 help = 'input dataset')
|
|
|
46 parser.add_argument('-n', '--none',
|
|
|
47 type = str,
|
|
|
48 default = 'true',
|
|
|
49 choices = ['true', 'false'],
|
|
|
50 help = 'compute Nan values')
|
|
|
51 parser.add_argument('-td', '--tool_dir',
|
|
|
52 type = str,
|
|
|
53 required = True,
|
|
|
54 help = 'your tool directory')
|
|
|
55 parser.add_argument('-na', '--name',
|
|
|
56 type = str,
|
|
|
57 help = 'name of dataset')
|
|
|
58 parser.add_argument('-de', '--dendro',
|
|
|
59 help = "Dendrogram out")
|
|
|
60 parser.add_argument('-ol', '--out_log',
|
|
|
61 help = "Output log")
|
|
|
62 parser.add_argument('-el', '--elbow',
|
|
|
63 help = "Out elbow")
|
|
|
64 args = parser.parse_args()
|
|
|
65 return args
|
|
|
66
|
|
|
67 ########################### warning ###########################################
|
|
|
68
|
|
|
69 def warning(s):
|
|
|
70 args = process_args(sys.argv)
|
|
|
71 with open(args.out_log, 'a') as log:
|
|
|
72 log.write(s)
|
|
|
73
|
|
|
74 ############################ dataset input ####################################
|
|
|
75
|
|
|
76 def read_dataset(data, name):
|
|
|
77 try:
|
|
|
78 dataset = pd.read_csv(data, sep = '\t', header = 0)
|
|
|
79 except pd.errors.EmptyDataError:
|
|
|
80 sys.exit('Execution aborted: wrong format of '+name+'\n')
|
|
|
81 if len(dataset.columns) < 2:
|
|
|
82 sys.exit('Execution aborted: wrong format of '+name+'\n')
|
|
|
83 return dataset
|
|
|
84
|
|
|
85 ############################ dataset name #####################################
|
|
|
86
|
|
|
87 def name_dataset(name_data, count):
|
|
|
88 if str(name_data) == 'Dataset':
|
|
|
89 return str(name_data) + '_' + str(count)
|
|
|
90 else:
|
|
|
91 return str(name_data)
|
|
|
92
|
|
|
93 ############################ load id e rules ##################################
|
|
|
94
|
|
|
95 def load_id_rules(reactions):
|
|
|
96 ids, rules = [], []
|
|
|
97 for key, value in reactions.items():
|
|
|
98 ids.append(key)
|
|
|
99 rules.append(value)
|
|
|
100 return (ids, rules)
|
|
|
101
|
|
|
102 ############################ check_methods ####################################
|
|
|
103
|
|
|
104 def gene_type(l, name):
|
|
|
105 if check_hgnc(l):
|
|
|
106 return 'hugo_id'
|
|
|
107 elif check_ensembl(l):
|
|
|
108 return 'ensembl_gene_id'
|
|
|
109 elif check_symbol(l):
|
|
|
110 return 'symbol'
|
|
|
111 elif check_entrez(l):
|
|
|
112 return 'entrez_id'
|
|
|
113 else:
|
|
|
114 sys.exit('Execution aborted:\n' +
|
|
|
115 'gene ID type in ' + name + ' not supported. Supported ID' +
|
|
|
116 'types are: HUGO ID, Ensemble ID, HUGO symbol, Entrez ID\n')
|
|
|
117
|
|
|
118 def check_hgnc(l):
|
|
|
119 if len(l) > 5:
|
|
|
120 if (l.upper()).startswith('HGNC:'):
|
|
|
121 return l[5:].isdigit()
|
|
|
122 else:
|
|
|
123 return False
|
|
|
124 else:
|
|
|
125 return False
|
|
|
126
|
|
|
127 def check_ensembl(l):
|
|
|
128 if len(l) == 15:
|
|
|
129 if (l.upper()).startswith('ENS'):
|
|
|
130 return l[4:].isdigit()
|
|
|
131 else:
|
|
|
132 return False
|
|
|
133 else:
|
|
|
134 return False
|
|
|
135
|
|
|
136 def check_symbol(l):
|
|
|
137 if len(l) > 0:
|
|
|
138 if l[0].isalpha() and l[1:].isalnum():
|
|
|
139 return True
|
|
|
140 else:
|
|
|
141 return False
|
|
|
142 else:
|
|
|
143 return False
|
|
|
144
|
|
|
145 def check_entrez(l):
|
|
|
146 if len(l) > 0:
|
|
|
147 return l.isdigit()
|
|
|
148 else:
|
|
|
149 return False
|
|
|
150
|
|
|
151 def check_bool(b):
|
|
|
152 if b == 'true':
|
|
|
153 return True
|
|
|
154 elif b == 'false':
|
|
|
155 return False
|
|
|
156
|
|
|
157 ############################ make recon #######################################
|
|
|
158
|
|
|
159 def check_and_doWord(l):
|
|
|
160 tmp = []
|
|
|
161 tmp_genes = []
|
|
|
162 count = 0
|
|
|
163 while l:
|
|
|
164 if count >= 0:
|
|
|
165 if l[0] == '(':
|
|
|
166 count += 1
|
|
|
167 tmp.append(l[0])
|
|
|
168 l.pop(0)
|
|
|
169 elif l[0] == ')':
|
|
|
170 count -= 1
|
|
|
171 tmp.append(l[0])
|
|
|
172 l.pop(0)
|
|
|
173 elif l[0] == ' ':
|
|
|
174 l.pop(0)
|
|
|
175 else:
|
|
|
176 word = []
|
|
|
177 while l:
|
|
|
178 if l[0] in [' ', '(', ')']:
|
|
|
179 break
|
|
|
180 else:
|
|
|
181 word.append(l[0])
|
|
|
182 l.pop(0)
|
|
|
183 word = ''.join(word)
|
|
|
184 tmp.append(word)
|
|
|
185 if not(word in ['or', 'and']):
|
|
|
186 tmp_genes.append(word)
|
|
|
187 else:
|
|
|
188 return False
|
|
|
189 if count == 0:
|
|
|
190 return (tmp, tmp_genes)
|
|
|
191 else:
|
|
|
192 return False
|
|
|
193
|
|
|
194 def brackets_to_list(l):
|
|
|
195 tmp = []
|
|
|
196 while l:
|
|
|
197 if l[0] == '(':
|
|
|
198 l.pop(0)
|
|
|
199 tmp.append(resolve_brackets(l))
|
|
|
200 else:
|
|
|
201 tmp.append(l[0])
|
|
|
202 l.pop(0)
|
|
|
203 return tmp
|
|
|
204
|
|
|
205 def resolve_brackets(l):
|
|
|
206 tmp = []
|
|
|
207 while l[0] != ')':
|
|
|
208 if l[0] == '(':
|
|
|
209 l.pop(0)
|
|
|
210 tmp.append(resolve_brackets(l))
|
|
|
211 else:
|
|
|
212 tmp.append(l[0])
|
|
|
213 l.pop(0)
|
|
|
214 l.pop(0)
|
|
|
215 return tmp
|
|
|
216
|
|
|
217 def priorityAND(l):
|
|
|
218 tmp = []
|
|
|
219 flag = True
|
|
|
220 while l:
|
|
|
221 if len(l) == 1:
|
|
|
222 if isinstance(l[0], list):
|
|
|
223 tmp.append(priorityAND(l[0]))
|
|
|
224 else:
|
|
|
225 tmp.append(l[0])
|
|
|
226 l = l[1:]
|
|
|
227 elif l[0] == 'or':
|
|
|
228 tmp.append(l[0])
|
|
|
229 flag = False
|
|
|
230 l = l[1:]
|
|
|
231 elif l[1] == 'or':
|
|
|
232 if isinstance(l[0], list):
|
|
|
233 tmp.append(priorityAND(l[0]))
|
|
|
234 else:
|
|
|
235 tmp.append(l[0])
|
|
|
236 tmp.append(l[1])
|
|
|
237 flag = False
|
|
|
238 l = l[2:]
|
|
|
239 elif l[1] == 'and':
|
|
|
240 tmpAnd = []
|
|
|
241 if isinstance(l[0], list):
|
|
|
242 tmpAnd.append(priorityAND(l[0]))
|
|
|
243 else:
|
|
|
244 tmpAnd.append(l[0])
|
|
|
245 tmpAnd.append(l[1])
|
|
|
246 if isinstance(l[2], list):
|
|
|
247 tmpAnd.append(priorityAND(l[2]))
|
|
|
248 else:
|
|
|
249 tmpAnd.append(l[2])
|
|
|
250 l = l[3:]
|
|
|
251 while l:
|
|
|
252 if l[0] == 'and':
|
|
|
253 tmpAnd.append(l[0])
|
|
|
254 if isinstance(l[1], list):
|
|
|
255 tmpAnd.append(priorityAND(l[1]))
|
|
|
256 else:
|
|
|
257 tmpAnd.append(l[1])
|
|
|
258 l = l[2:]
|
|
|
259 elif l[0] == 'or':
|
|
|
260 flag = False
|
|
|
261 break
|
|
|
262 if flag == True: #se ci sono solo AND nella lista
|
|
|
263 tmp.extend(tmpAnd)
|
|
|
264 elif flag == False:
|
|
|
265 tmp.append(tmpAnd)
|
|
|
266 return tmp
|
|
|
267
|
|
|
268 def checkRule(l):
|
|
|
269 if len(l) == 1:
|
|
|
270 if isinstance(l[0], list):
|
|
|
271 if checkRule(l[0]) is False:
|
|
|
272 return False
|
|
|
273 elif len(l) > 2:
|
|
|
274 if checkRule2(l) is False:
|
|
|
275 return False
|
|
|
276 else:
|
|
|
277 return False
|
|
|
278 return True
|
|
|
279
|
|
|
280 def checkRule2(l):
|
|
|
281 while l:
|
|
|
282 if len(l) == 1:
|
|
|
283 return False
|
|
|
284 elif isinstance(l[0], list) and l[1] in ['and', 'or']:
|
|
|
285 if checkRule(l[0]) is False:
|
|
|
286 return False
|
|
|
287 if isinstance(l[2], list):
|
|
|
288 if checkRule(l[2]) is False:
|
|
|
289 return False
|
|
|
290 l = l[3:]
|
|
|
291 elif l[1] in ['and', 'or']:
|
|
|
292 if isinstance(l[2], list):
|
|
|
293 if checkRule(l[2]) is False:
|
|
|
294 return False
|
|
|
295 l = l[3:]
|
|
|
296 elif l[0] in ['and', 'or']:
|
|
|
297 if isinstance(l[1], list):
|
|
|
298 if checkRule(l[1]) is False:
|
|
|
299 return False
|
|
|
300 l = l[2:]
|
|
|
301 else:
|
|
|
302 return False
|
|
|
303 return True
|
|
|
304
|
|
|
305 def do_rules(rules):
|
|
|
306 split_rules = []
|
|
|
307 err_rules = []
|
|
|
308 tmp_gene_in_rule = []
|
|
|
309 for i in range(len(rules)):
|
|
|
310 tmp = list(rules[i])
|
|
|
311 if tmp:
|
|
|
312 tmp, tmp_genes = check_and_doWord(tmp)
|
|
|
313 tmp_gene_in_rule.extend(tmp_genes)
|
|
|
314 if tmp is False:
|
|
|
315 split_rules.append([])
|
|
|
316 err_rules.append(rules[i])
|
|
|
317 else:
|
|
|
318 tmp = brackets_to_list(tmp)
|
|
|
319 if checkRule(tmp):
|
|
|
320 split_rules.append(priorityAND(tmp))
|
|
|
321 else:
|
|
|
322 split_rules.append([])
|
|
|
323 err_rules.append(rules[i])
|
|
|
324 else:
|
|
|
325 split_rules.append([])
|
|
|
326 if err_rules:
|
|
|
327 warning('Warning: wrong format rule in ' + str(err_rules) + '\n')
|
|
|
328 return (split_rules, list(set(tmp_gene_in_rule)))
|
|
|
329
|
|
|
330 def make_recon(data):
|
|
|
331 try:
|
|
|
332 import cobra as cb
|
|
|
333 import warnings
|
|
|
334 with warnings.catch_warnings():
|
|
|
335 warnings.simplefilter('ignore')
|
|
|
336 recon = cb.io.read_sbml_model(data)
|
|
|
337 react = recon.reactions
|
|
|
338 rules = [react[i].gene_reaction_rule for i in range(len(react))]
|
|
|
339 ids = [react[i].id for i in range(len(react))]
|
|
|
340 except cb.io.sbml3.CobraSBMLError:
|
|
|
341 try:
|
|
|
342 data = (pd.read_csv(data, sep = '\t', dtype = str)).fillna('')
|
|
|
343 if len(data.columns) < 2:
|
|
|
344 sys.exit('Execution aborted: wrong format of ' +
|
|
|
345 'custom GPR rules\n')
|
|
|
346 if not len(data.columns) == 2:
|
|
|
347 warning('WARNING: more than 2 columns in custom GPR rules.\n' +
|
|
|
348 'Extra columns have been disregarded\n')
|
|
|
349 ids = list(data.iloc[:, 0])
|
|
|
350 rules = list(data.iloc[:, 1])
|
|
|
351 except pd.errors.EmptyDataError:
|
|
|
352 sys.exit('Execution aborted: wrong format of custom GPR rules\n')
|
|
|
353 except pd.errors.ParserError:
|
|
|
354 sys.exit('Execution aborted: wrong format of custom GPR rules\n')
|
|
|
355 split_rules, tmp_genes = do_rules(rules)
|
|
|
356 gene_in_rule = {}
|
|
|
357 for i in tmp_genes:
|
|
|
358 gene_in_rule[i] = 'ok'
|
|
|
359 return (ids, split_rules, gene_in_rule)
|
|
|
360
|
|
|
361 ############################ resolve_methods ##################################
|
|
|
362
|
|
|
363 def replace_gene_value(l, d):
|
|
|
364 tmp = []
|
|
|
365 err = []
|
|
|
366 while l:
|
|
|
367 if isinstance(l[0], list):
|
|
|
368 tmp_rules, tmp_err = replace_gene_value(l[0], d)
|
|
|
369 tmp.append(tmp_rules)
|
|
|
370 err.extend(tmp_err)
|
|
|
371 else:
|
|
|
372 value = replace_gene(l[0],d)
|
|
|
373 tmp.append(value)
|
|
|
374 if value == None:
|
|
|
375 err.append(l[0])
|
|
|
376 l = l[1:]
|
|
|
377 return (tmp, err)
|
|
|
378
|
|
|
379 def replace_gene(l, d):
|
|
|
380 if l =='and' or l == 'or':
|
|
|
381 return l
|
|
|
382 else:
|
|
|
383 value = d.get(l, None)
|
|
|
384 if not(value == None or isinstance(value, (int, float))):
|
|
|
385 sys.exit('Execution aborted: ' + value + ' value not valid\n')
|
|
|
386 return value
|
|
|
387
|
|
|
388 def compute(val1, op, val2, cn):
|
|
|
389 if val1 != None and val2 != None:
|
|
|
390 if op == 'and':
|
|
|
391 return min(val1, val2)
|
|
|
392 else:
|
|
|
393 return val1 + val2
|
|
|
394 elif op == 'and':
|
|
|
395 if cn is True:
|
|
|
396 if val1 != None:
|
|
|
397 return val1
|
|
|
398 elif val2 != None:
|
|
|
399 return val2
|
|
|
400 else:
|
|
|
401 return None
|
|
|
402 else:
|
|
|
403 return None
|
|
|
404 else:
|
|
|
405 if val1 != None:
|
|
|
406 return val1
|
|
|
407 elif val2 != None:
|
|
|
408 return val2
|
|
|
409 else:
|
|
|
410 return None
|
|
|
411
|
|
|
412 def control(ris, l, cn):
|
|
|
413 if len(l) == 1:
|
|
|
414 if isinstance(l[0], (float, int)) or l[0] == None:
|
|
|
415 return l[0]
|
|
|
416 elif isinstance(l[0], list):
|
|
|
417 return control(None, l[0], cn)
|
|
|
418 else:
|
|
|
419 return False
|
|
|
420 elif len(l) > 2:
|
|
|
421 return control_list(ris, l, cn)
|
|
|
422 else:
|
|
|
423 return False
|
|
|
424
|
|
|
425 def control_list(ris, l, cn):
|
|
|
426 while l:
|
|
|
427 if len(l) == 1:
|
|
|
428 return False
|
|
|
429 elif (isinstance(l[0], (float, int)) or
|
|
|
430 l[0] == None) and l[1] in ['and', 'or']:
|
|
|
431 if isinstance(l[2], (float, int)) or l[2] == None:
|
|
|
432 ris = compute(l[0], l[1], l[2], cn)
|
|
|
433 elif isinstance(l[2], list):
|
|
|
434 tmp = control(None, l[2], cn)
|
|
|
435 if tmp is False:
|
|
|
436 return False
|
|
|
437 else:
|
|
|
438 ris = compute(l[0], l[1], tmp, cn)
|
|
|
439 else:
|
|
|
440 return False
|
|
|
441 l = l[3:]
|
|
|
442 elif l[0] in ['and', 'or']:
|
|
|
443 if isinstance(l[1], (float, int)) or l[1] == None:
|
|
|
444 ris = compute(ris, l[0], l[1], cn)
|
|
|
445 elif isinstance(l[1], list):
|
|
|
446 tmp = control(None,l[1], cn)
|
|
|
447 if tmp is False:
|
|
|
448 return False
|
|
|
449 else:
|
|
|
450 ris = compute(ris, l[0], tmp, cn)
|
|
|
451 else:
|
|
|
452 return False
|
|
|
453 l = l[2:]
|
|
|
454 elif isinstance(l[0], list) and l[1] in ['and', 'or']:
|
|
|
455 if isinstance(l[2], (float, int)) or l[2] == None:
|
|
|
456 tmp = control(None, l[0], cn)
|
|
|
457 if tmp is False:
|
|
|
458 return False
|
|
|
459 else:
|
|
|
460 ris = compute(tmp, l[1], l[2], cn)
|
|
|
461 elif isinstance(l[2], list):
|
|
|
462 tmp = control(None, l[0], cn)
|
|
|
463 tmp2 = control(None, l[2], cn)
|
|
|
464 if tmp is False or tmp2 is False:
|
|
|
465 return False
|
|
|
466 else:
|
|
|
467 ris = compute(tmp, l[1], tmp2, cn)
|
|
|
468 else:
|
|
|
469 return False
|
|
|
470 l = l[3:]
|
|
|
471 else:
|
|
|
472 return False
|
|
|
473 return ris
|
|
|
474
|
|
|
475 ############################ gene #############################################
|
|
|
476
|
|
|
477 def data_gene(gene, type_gene, name, gene_custom):
|
|
|
478 args = process_args(sys.argv)
|
|
|
479 for i in range(len(gene)):
|
|
|
480 tmp = gene.iloc[i, 0]
|
|
|
481 if tmp.startswith(' ') or tmp.endswith(' '):
|
|
|
482 gene.iloc[i, 0] = (tmp.lstrip()).rstrip()
|
|
|
483 gene_dup = [item for item, count in
|
|
|
484 collections.Counter(gene[gene.columns[0]]).items() if count > 1]
|
|
|
485 pat_dup = [item for item, count in
|
|
|
486 collections.Counter(list(gene.columns)).items() if count > 1]
|
|
|
487 if gene_dup:
|
|
|
488 if gene_custom == None:
|
|
|
489 if args.rules_selector == 'HMRcore':
|
|
|
490 gene_in_rule = pk.load(open(args.tool_dir +
|
|
|
491 '/local/HMRcore_genes.p', 'rb'))
|
|
|
492 elif args.rules_selector == 'Recon':
|
|
|
493 gene_in_rule = pk.load(open(args.tool_dir +
|
|
|
494 '/local/Recon_genes.p', 'rb'))
|
|
|
495 gene_in_rule = gene_in_rule.get(type_gene)
|
|
|
496 else:
|
|
|
497 gene_in_rule = gene_custom
|
|
|
498 tmp = []
|
|
|
499 for i in gene_dup:
|
|
|
500 if gene_in_rule.get(i) == 'ok':
|
|
|
501 tmp.append(i)
|
|
|
502 if tmp:
|
|
|
503 sys.exit('Execution aborted because gene ID '
|
|
|
504 + str(tmp) + ' in ' + name + ' is duplicated\n')
|
|
|
505 if pat_dup:
|
|
|
506 sys.exit('Execution aborted: duplicated label\n'
|
|
|
507 + str(pat_dup) + 'in ' + name + '\n')
|
|
|
508 return (gene.set_index(gene.columns[0])).to_dict()
|
|
|
509
|
|
|
510 ############################ resolve ##########################################
|
|
|
511
|
|
|
512 def resolve(genes, rules, ids, resolve_none, name):
|
|
|
513 resolve_rules = {}
|
|
|
514 not_found = []
|
|
|
515 flag = False
|
|
|
516 for key, value in genes.items():
|
|
|
517 tmp_resolve = []
|
|
|
518 for i in range(len(rules)):
|
|
|
519 tmp = rules[i]
|
|
|
520 if tmp:
|
|
|
521 tmp, err = replace_gene_value(tmp, value)
|
|
|
522 if err:
|
|
|
523 not_found.extend(err)
|
|
|
524 ris = control(None, tmp, resolve_none)
|
|
|
525 if ris is False or ris == None:
|
|
|
526 tmp_resolve.append(None)
|
|
|
527 else:
|
|
|
528 tmp_resolve.append(ris)
|
|
|
529 flag = True
|
|
|
530 else:
|
|
|
531 tmp_resolve.append(None)
|
|
|
532 resolve_rules[key] = tmp_resolve
|
|
|
533 if flag is False:
|
|
|
534 sys.exit('Execution aborted: no computable score' +
|
|
|
535 ' (due to missing gene values) for class '
|
|
|
536 + name + ', the class has been disregarded\n')
|
|
|
537 return (resolve_rules, list(set(not_found)))
|
|
|
538
|
|
|
539 ################################# clustering ##################################
|
|
|
540
|
|
|
541 def f_cluster(resolve_rules):
|
|
|
542 os.makedirs('cluster_out')
|
|
|
543 args = process_args(sys.argv)
|
|
13
|
544 k_min = args.k_min
|
|
|
545 k_max = args.k_max
|
|
|
546 if k_min > k_max:
|
|
|
547 warning('k range boundaries inverted.\n')
|
|
|
548 tmp = k_min
|
|
|
549 k_min = k_max
|
|
|
550 k_max = tmp
|
|
|
551 else:
|
|
|
552 warning('k range correct.\n')
|
|
0
|
553 cluster_data = pd.DataFrame.from_dict(resolve_rules, orient = 'index')
|
|
|
554 for i in cluster_data.columns:
|
|
|
555 tmp = cluster_data[i][0]
|
|
|
556 if tmp == None:
|
|
|
557 cluster_data = cluster_data.drop(columns=[i])
|
|
|
558 distorsion = []
|
|
13
|
559 for i in range(k_min, k_max+1):
|
|
0
|
560 tmp_kmeans = KMeans(n_clusters = i,
|
|
|
561 n_init = 100,
|
|
|
562 max_iter = 300,
|
|
|
563 random_state = 0).fit(cluster_data)
|
|
|
564 distorsion.append(tmp_kmeans.inertia_)
|
|
|
565 predict = tmp_kmeans.predict(cluster_data)
|
|
|
566 predict = [x+1 for x in predict]
|
|
7
|
567 classe = (pd.DataFrame(list(zip(cluster_data.index, predict)))).astype(str)
|
|
0
|
568 dest = 'cluster_out/K=' + str(i) + '_' + args.name+'.tsv'
|
|
|
569 classe.to_csv(dest, sep = '\t', index = False,
|
|
|
570 header = ['Patient_ID', 'Class'])
|
|
|
571 plt.figure(0)
|
|
13
|
572 plt.plot(range(k_min, k_max+1), distorsion, marker = 'o')
|
|
0
|
573 plt.xlabel('Number of cluster')
|
|
|
574 plt.ylabel('Distorsion')
|
|
|
575 plt.savefig(args.elbow, dpi = 240, format = 'pdf')
|
|
|
576 if args.cond_hier == 'yes':
|
|
|
577 import scipy.cluster.hierarchy as hier
|
|
|
578 lin = hier.linkage(cluster_data, args.linkage)
|
|
|
579 plt.figure(1)
|
|
|
580 plt.figure(figsize=(10, 5))
|
|
|
581 hier.dendrogram(lin, leaf_font_size = 2, labels = cluster_data.index)
|
|
|
582 plt.savefig(args.dendro, dpi = 480, format = 'pdf')
|
|
|
583 return None
|
|
|
584
|
|
|
585 ################################# main ########################################
|
|
|
586
|
|
|
587 def main():
|
|
|
588 args = process_args(sys.argv)
|
|
|
589 if args.rules_selector == 'HMRcore':
|
|
|
590 recon = pk.load(open(args.tool_dir + '/local/HMRcore_rules.p', 'rb'))
|
|
|
591 elif args.rules_selector == 'Recon':
|
|
|
592 recon = pk.load(open(args.tool_dir + '/local/Recon_rules.p', 'rb'))
|
|
|
593 elif args.rules_selector == 'Custom':
|
|
|
594 ids, rules, gene_in_rule = make_recon(args.custom)
|
|
|
595 resolve_none = check_bool(args.none)
|
|
|
596 dataset = read_dataset(args.data, args.name)
|
|
|
597 dataset.iloc[:, 0] = (dataset.iloc[:, 0]).astype(str)
|
|
|
598 type_gene = gene_type(dataset.iloc[0, 0], args.name)
|
|
|
599 if args.rules_selector != 'Custom':
|
|
|
600 genes = data_gene(dataset, type_gene, args.name, None)
|
|
|
601 ids, rules = load_id_rules(recon.get(type_gene))
|
|
|
602 elif args.rules_selector == 'Custom':
|
|
|
603 genes = data_gene(dataset, type_gene, args.name, gene_in_rule)
|
|
|
604 resolve_rules, err = resolve(genes, rules, ids, resolve_none, args.name)
|
|
|
605 if err:
|
|
|
606 warning('WARNING: gene\n' + str(err) + '\nnot found in class '
|
|
|
607 + args.name + ', the expression level for this gene ' +
|
|
|
608 'will be considered NaN\n')
|
|
|
609 f_cluster(resolve_rules)
|
|
|
610 warning('Execution succeeded')
|
|
|
611 return None
|
|
|
612
|
|
|
613 ###############################################################################
|
|
|
614
|
|
|
615 if __name__ == "__main__":
|
|
6
|
616 main()
|
