Mercurial > repos > bimib > marea

comparison Marea/marea_cluster.py @ 16:c71ac0bb12de draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author bimib
date Tue, 01 Oct 2019 06:05:13 -0400
parents 1a0c8c2780f2
children a8825e66c3a0
comparison
equal deleted inserted replaced
15:d0e7f14b773f 16:c71ac0bb12de
1 from __future__ import division 1 # -*- coding: utf-8 -*-
2 """
3 Created on Mon Jun 3 19:51:00 2019
4
5 @author: Narger
6 """
7
8 import sys
9 import argparse
2 import os 10 import os
3 import sys 11 from sklearn.datasets import make_blobs
12 from sklearn.cluster import KMeans, DBSCAN, AgglomerativeClustering
13 from sklearn.metrics import silhouette_samples, silhouette_score, davies_bouldin_score, cluster
14 import matplotlib.pyplot as plt
15 import scipy.cluster.hierarchy as shc
16 import matplotlib.cm as cm
17 import numpy as np
4 import pandas as pd 18 import pandas as pd
5 import collections 19
6 import pickle as pk 20 ################################# process args ###############################
7 import argparse
8 from sklearn.cluster import KMeans
9 import matplotlib
10 # Force matplotlib to not use any Xwindows backend.
11 matplotlib.use('Agg')
12 import matplotlib.pyplot as plt
13
14 ########################## argparse ###########################################
15 21
16 def process_args(args): 22 def process_args(args):
17 parser = argparse.ArgumentParser(usage = '%(prog)s [options]', 23 parser = argparse.ArgumentParser(usage = '%(prog)s [options]',
18 description = 'process some value\'s' + 24 description = 'process some value\'s' +
19 ' genes to create class.') 25 ' genes to create class.')
20 parser.add_argument('-rs', '--rules_selector', 26
21 type = str, 27 parser.add_argument('-ol', '--out_log',
22 default = 'HMRcore', 28 help = "Output log")
23 choices = ['HMRcore', 'Recon', 'Custom'], 29
24 help = 'chose which type of dataset you want use') 30 parser.add_argument('-in', '--input',
25 parser.add_argument('-cr', '--custom', 31 type = str,
26 type = str, 32 help = 'input dataset')
27 help='your dataset if you want custom rules') 33
28 parser.add_argument('-ch', '--cond_hier', 34 parser.add_argument('-cy', '--cluster_type',
29 type = str, 35 type = str,
30 default = 'no', 36 choices = ['kmeans', 'meanshift', 'dbscan', 'hierarchy'],
31 choices = ['no', 'yes'], 37 default = 'kmeans',
32 help = 'chose if you wanna hierical dendrogram') 38 help = 'choose clustering algorythm')
33 parser.add_argument('-lk', '--k_min', 39
40 parser.add_argument('-k1', '--k_min',
34 type = int, 41 type = int,
35 help = 'min number of cluster') 42 default = 2,
36 parser.add_argument('-uk', '--k_max', 43 help = 'choose minimun cluster number to be generated')
44
45 parser.add_argument('-k2', '--k_max',
37 type = int, 46 type = int,
38 help = 'max number of cluster') 47 default = 7,
39 parser.add_argument('-li', '--linkage', 48 help = 'choose maximum cluster number to be generated')
40 type = str, 49
41 choices = ['single', 'complete', 'average'], 50 parser.add_argument('-el', '--elbow',
42 help='linkage hierarchical cluster') 51 type = str,
43 parser.add_argument('-d', '--data', 52 default = 'false',
44 type = str, 53 choices = ['true', 'false'],
45 required = True, 54 help = 'choose if you want to generate an elbow plot for kmeans')
46 help = 'input dataset') 55
47 parser.add_argument('-n', '--none', 56 parser.add_argument('-si', '--silhouette',
48 type = str, 57 type = str,
49 default = 'true', 58 default = 'false',
50 choices = ['true', 'false'], 59 choices = ['true', 'false'],
51 help = 'compute Nan values') 60 help = 'choose if you want silhouette plots')
61
62 parser.add_argument('-db', '--davies',
63 type = str,
64 default = 'false',
65 choices = ['true', 'false'],
66 help = 'choose if you want davies bouldin scores')
67
52 parser.add_argument('-td', '--tool_dir', 68 parser.add_argument('-td', '--tool_dir',
53 type = str, 69 type = str,
54 required = True, 70 required = True,
55 help = 'your tool directory') 71 help = 'your tool directory')
56 parser.add_argument('-na', '--name', 72
57 type = str, 73 parser.add_argument('-ms', '--min_samples',
58 help = 'name of dataset') 74 type = int,
59 parser.add_argument('-de', '--dendro', 75 help = 'min samples for dbscan (optional)')
60 help = "Dendrogram out") 76
61 parser.add_argument('-ol', '--out_log', 77 parser.add_argument('-ep', '--eps',
62 help = "Output log") 78 type = int,
63 parser.add_argument('-el', '--elbow', 79 help = 'eps for dbscan (optional)')
64 help = "Out elbow") 80
81
65 args = parser.parse_args() 82 args = parser.parse_args()
66 return args 83 return args
67 84
68 ########################### warning ########################################### 85 ########################### warning ###########################################
69 86
70 def warning(s): 87 def warning(s):
71 args = process_args(sys.argv) 88 args = process_args(sys.argv)
72 with open(args.out_log, 'a') as log: 89 with open(args.out_log, 'a') as log:
73 log.write(s) 90 log.write(s + "\n\n")
74 91 print(s)
75 ############################ dataset input #################################### 92
76 93 ########################## read dataset ######################################
77 def read_dataset(data, name): 94
95 def read_dataset(dataset):
78 try: 96 try:
79 dataset = pd.read_csv(data, sep = '\t', header = 0) 97 dataset = pd.read_csv(dataset, sep = '\t', header = 0)
80 except pd.errors.EmptyDataError: 98 except pd.errors.EmptyDataError:
81 sys.exit('Execution aborted: wrong format of '+name+'\n') 99 sys.exit('Execution aborted: wrong format of dataset\n')
82 if len(dataset.columns) < 2: 100 if len(dataset.columns) < 2:
83 sys.exit('Execution aborted: wrong format of '+name+'\n') 101 sys.exit('Execution aborted: wrong format of dataset\n')
84 return dataset 102 return dataset
85 103
86 ############################ dataset name ##################################### 104 ############################ rewrite_input ###################################
87 105
88 def name_dataset(name_data, count): 106 def rewrite_input(dataset):
89 if str(name_data) == 'Dataset': 107 #Riscrivo il dataset come dizionario di liste,
90 return str(name_data) + '_' + str(count) 108 #non come dizionario di dizionari
109
110 for key, val in dataset.items():
111 l = []
112 for i in val:
113 if i == 'None':
114 l.append(None)
115 else:
116 l.append(float(i))
117
118 dataset[key] = l
119
120 return dataset
121
122 ############################## write to csv ##################################
123
124 def write_to_csv (dataset, labels, name):
125 list_labels = labels
126 list_values = dataset
127
128 list_values = list_values.tolist()
129 d = {'Label' : list_labels, 'Value' : list_values}
130
131 df = pd.DataFrame(d, columns=['Value','Label'])
132
133 dest = name + '.tsv'
134 df.to_csv(dest, sep = '\t', index = False,
135 header = ['Value', 'Label'])
136
137 ########################### trova il massimo in lista ########################
138 def max_index (lista):
139 best = -1
140 best_index = 0
141 for i in range(len(lista)):
142 if lista[i] > best:
143 best = lista [i]
144 best_index = i
145
146 return best_index
147
148 ################################ kmeans #####################################
149
150 def kmeans (k_min, k_max, dataset, elbow, silhouette, davies):
151 if not os.path.exists('clustering/kmeans_output'):
152 os.makedirs('clustering/kmeans_output')
153
154
155 if elbow == 'true':
156 elbow = True
91 else: 157 else:
92 return str(name_data) 158 elbow = False
93 159
94 ############################ load id e rules ################################## 160 if silhouette == 'true':
95 161 silhouette = True
96 def load_id_rules(reactions):
97 ids, rules = [], []
98 for key, value in reactions.items():
99 ids.append(key)
100 rules.append(value)
101 return (ids, rules)
102
103 ############################ check_methods ####################################
104
105 def gene_type(l, name):
106 if check_hgnc(l):
107 return 'hugo_id'
108 elif check_ensembl(l):
109 return 'ensembl_gene_id'
110 elif check_symbol(l):
111 return 'symbol'
112 elif check_entrez(l):
113 return 'entrez_id'
114 else: 162 else:
115 sys.exit('Execution aborted:\n' + 163 silhouette = False
116 'gene ID type in ' + name + ' not supported. Supported ID' + 164
117 'types are: HUGO ID, Ensemble ID, HUGO symbol, Entrez ID\n') 165 if davies == 'true':
118 166 davies = True
119 def check_hgnc(l):
120 if len(l) > 5:
121 if (l.upper()).startswith('HGNC:'):
122 return l[5:].isdigit()
123 else:
124 return False
125 else: 167 else:
126 return False 168 davies = False
127 169
128 def check_ensembl(l): 170
129 if len(l) == 15: 171 range_n_clusters = [i for i in range(k_min, k_max+1)]
130 if (l.upper()).startswith('ENS'): 172 distortions = []
131 return l[4:].isdigit() 173 scores = []
132 else: 174 all_labels = []
133 return False 175
134 else: 176 for n_clusters in range_n_clusters:
135 return False 177 clusterer = KMeans(n_clusters=n_clusters, random_state=10)
136 178 cluster_labels = clusterer.fit_predict(dataset)
137 def check_symbol(l): 179
138 if len(l) > 0: 180 all_labels.append(cluster_labels)
139 if l[0].isalpha() and l[1:].isalnum(): 181 silhouette_avg = silhouette_score(dataset, cluster_labels)
140 return True 182 scores.append(silhouette_avg)
141 else: 183 distortions.append(clusterer.fit(dataset).inertia_)
142 return False 184
185 best = max_index(scores) + k_min
186
187 for i in range(len(all_labels)):
188 prefix = ''
189 if (i + k_min == best):
190 prefix = '_BEST'
191
192 write_to_csv(dataset, all_labels[i], 'clustering/kmeans_output/kmeans_with_' + str(i + k_min) + prefix + '_clusters.tsv')
193
194 if davies:
195 with np.errstate(divide='ignore', invalid='ignore'):
196 davies_bouldin = davies_bouldin_score(dataset, all_labels[i])
197 warning("\nFor n_clusters = " + str(i + k_min) +
198 " The average davies bouldin score is: " + str(davies_bouldin))
199
200
201 if silhouette:
202 silihouette_draw(dataset, all_labels[i], i + k_min, 'clustering/kmeans_output/silhouette_with_' + str(i + k_min) + prefix + '_clusters.png')
203
204
205 if elbow:
206 elbow_plot(distortions, k_min,k_max)
207
208
209
210
211
212 ############################## elbow_plot ####################################
213
214 def elbow_plot (distortions, k_min, k_max):
215 plt.figure(0)
216 plt.plot(range(k_min, k_max+1), distortions, marker = 'o')
217 plt.xlabel('Number of cluster')
218 plt.ylabel('Distortion')
219 s = 'clustering/kmeans_output/elbow_plot.png'
220 fig = plt.gcf()
221 fig.set_size_inches(18.5, 10.5, forward = True)
222 fig.savefig(s, dpi=100)
223
224
225 ############################## silhouette plot ###############################
226 def silihouette_draw(dataset, labels, n_clusters, path):
227 silhouette_avg = silhouette_score(dataset, labels)
228 warning("For n_clusters = " + str(n_clusters) +
229 " The average silhouette_score is: " + str(silhouette_avg))
230
231 plt.close('all')
232 # Create a subplot with 1 row and 2 columns
233 fig, (ax1) = plt.subplots(1, 1)
234
235 fig.set_size_inches(18, 7)
236
237 # The 1st subplot is the silhouette plot
238 # The silhouette coefficient can range from -1, 1 but in this example all
239 # lie within [-0.1, 1]
240 ax1.set_xlim([-1, 1])
241 # The (n_clusters+1)*10 is for inserting blank space between silhouette
242 # plots of individual clusters, to demarcate them clearly.
243 ax1.set_ylim([0, len(dataset) + (n_clusters + 1) * 10])
244
245 # Compute the silhouette scores for each sample
246 sample_silhouette_values = silhouette_samples(dataset, labels)
247
248 y_lower = 10
249 for i in range(n_clusters):
250 # Aggregate the silhouette scores for samples belonging to
251 # cluster i, and sort them
252 ith_cluster_silhouette_values = \
253 sample_silhouette_values[labels == i]
254
255 ith_cluster_silhouette_values.sort()
256
257 size_cluster_i = ith_cluster_silhouette_values.shape[0]
258 y_upper = y_lower + size_cluster_i
259
260 color = cm.nipy_spectral(float(i) / n_clusters)
261 ax1.fill_betweenx(np.arange(y_lower, y_upper),
262 0, ith_cluster_silhouette_values,
263 facecolor=color, edgecolor=color, alpha=0.7)
264
265 # Label the silhouette plots with their cluster numbers at the middle
266 ax1.text(-0.05, y_lower + 0.5 * size_cluster_i, str(i))
267
268 # Compute the new y_lower for next plot
269 y_lower = y_upper + 10 # 10 for the 0 samples
270
271 ax1.set_title("The silhouette plot for the various clusters.")
272 ax1.set_xlabel("The silhouette coefficient values")
273 ax1.set_ylabel("Cluster label")
274
275 # The vertical line for average silhouette score of all the values
276 ax1.axvline(x=silhouette_avg, color="red", linestyle="--")
277
278 ax1.set_yticks([]) # Clear the yaxis labels / ticks
279 ax1.set_xticks([-0.1, 0, 0.2, 0.4, 0.6, 0.8, 1])
280
281
282 plt.suptitle(("Silhouette analysis for clustering on sample data "
283 "with n_clusters = " + str(n_clusters) + "\nAverage silhouette_score = " + str(silhouette_avg)), fontsize=12, fontweight='bold')
284
285
286 plt.savefig(path, bbox_inches='tight')
287
288 ######################## dbscan ##############################################
289
290 def dbscan(dataset, eps, min_samples):
291 if not os.path.exists('clustering/dbscan_output'):
292 os.makedirs('clustering/dbscan_output')
293
294 if eps is not None:
295 clusterer = DBSCAN(eps = eps, min_samples = min_samples)
143 else: 296 else:
144 return False 297 clusterer = DBSCAN()
145 298
146 def check_entrez(l): 299 clustering = clusterer.fit(dataset)
147 if len(l) > 0: 300
148 return l.isdigit() 301 core_samples_mask = np.zeros_like(clustering.labels_, dtype=bool)
149 else: 302 core_samples_mask[clustering.core_sample_indices_] = True
150 return False 303 labels = clustering.labels_
151 304
152 def check_bool(b): 305 # Number of clusters in labels, ignoring noise if present.
153 if b == 'true': 306 n_clusters_ = len(set(labels)) - (1 if -1 in labels else 0)
154 return True 307
155 elif b == 'false': 308 silhouette_avg = silhouette_score(dataset, labels)
156 return False 309 warning("For n_clusters =" + str(n_clusters_) +
157 310 "The average silhouette_score is :" + str(silhouette_avg))
158 ############################ make recon ####################################### 311
159 312 ##TODO: PLOT SU DBSCAN (no centers) e HIERARCHICAL
160 def check_and_doWord(l): 313
161 tmp = [] 314 # Black removed and is used for noise instead.
162 tmp_genes = [] 315 unique_labels = set(labels)
163 count = 0 316 colors = [plt.cm.Spectral(each)
164 while l: 317 for each in np.linspace(0, 1, len(unique_labels))]
165 if count >= 0: 318 for k, col in zip(unique_labels, colors):
166 if l[0] == '(': 319 if k == -1:
167 count += 1 320 # Black used for noise.
168 tmp.append(l[0]) 321 col = [0, 0, 0, 1]
169 l.pop(0) 322
170 elif l[0] == ')': 323 class_member_mask = (labels == k)
171 count -= 1 324
172 tmp.append(l[0]) 325 xy = dataset[class_member_mask & core_samples_mask]
173 l.pop(0) 326 plt.plot(xy[:, 0], xy[:, 1], 'o', markerfacecolor=tuple(col),
174 elif l[0] == ' ': 327 markeredgecolor='k', markersize=14)
175 l.pop(0) 328
176 else: 329 xy = dataset[class_member_mask & ~core_samples_mask]
177 word = [] 330 plt.plot(xy[:, 0], xy[:, 1], 'o', markerfacecolor=tuple(col),
178 while l: 331 markeredgecolor='k', markersize=6)
179 if l[0] in [' ', '(', ')']: 332
180 break 333 plt.title('Estimated number of clusters: %d' % n_clusters_)
181 else: 334 s = 'clustering/dbscan_output/dbscan_plot.png'
182 word.append(l[0]) 335 fig = plt.gcf()
183 l.pop(0) 336 fig.set_size_inches(18.5, 10.5, forward = True)
184 word = ''.join(word) 337 fig.savefig(s, dpi=100)
185 tmp.append(word) 338
186 if not(word in ['or', 'and']): 339
187 tmp_genes.append(word) 340 write_to_csv(dataset, labels, 'clustering/dbscan_output/dbscan_results.tsv')
188 else: 341
189 return False 342 ########################## hierachical #######################################
190 if count == 0: 343
191 return (tmp, tmp_genes) 344 def hierachical_agglomerative(dataset, k_min, k_max):
192 else: 345
193 return False 346 if not os.path.exists('clustering/agglomerative_output'):
194 347 os.makedirs('clustering/agglomerative_output')
195 def brackets_to_list(l): 348
196 tmp = [] 349 plt.figure(figsize=(10, 7))
197 while l: 350 plt.title("Customer Dendograms")
198 if l[0] == '(': 351 shc.dendrogram(shc.linkage(dataset, method='ward'))
199 l.pop(0) 352 fig = plt.gcf()
200 tmp.append(resolve_brackets(l)) 353 fig.savefig('clustering/agglomerative_output/dendogram.png', dpi=200)
201 else: 354
202 tmp.append(l[0]) 355 range_n_clusters = [i for i in range(k_min, k_max+1)]
203 l.pop(0) 356
204 return tmp 357 for n_clusters in range_n_clusters:
205 358
206 def resolve_brackets(l): 359 cluster = AgglomerativeClustering(n_clusters=n_clusters, affinity='euclidean', linkage='ward')
207 tmp = [] 360 cluster.fit_predict(dataset)
208 while l[0] != ')': 361 cluster_labels = cluster.labels_
209 if l[0] == '(': 362
210 l.pop(0) 363 silhouette_avg = silhouette_score(dataset, cluster_labels)
211 tmp.append(resolve_brackets(l)) 364 warning("For n_clusters =", n_clusters,
212 else: 365 "The average silhouette_score is :", silhouette_avg)
213 tmp.append(l[0]) 366
214 l.pop(0) 367 plt.clf()
215 l.pop(0) 368 plt.figure(figsize=(10, 7))
216 return tmp 369 plt.title("Agglomerative Hierarchical Clustering\nwith " + str(n_clusters) + " clusters and " + str(silhouette_avg) + " silhouette score")
217 370 plt.scatter(dataset[:,0], dataset[:,1], c = cluster_labels, cmap='rainbow')
218 def priorityAND(l): 371 s = 'clustering/agglomerative_output/hierachical_' + str(n_clusters) + '_clusters.png'
219 tmp = [] 372 fig = plt.gcf()
220 flag = True 373 fig.set_size_inches(10, 7, forward = True)
221 while l: 374 fig.savefig(s, dpi=200)
222 if len(l) == 1: 375
223 if isinstance(l[0], list): 376 write_to_csv(dataset, cluster_labels, 'clustering/agglomerative_output/agglomerative_hierarchical_with_' + str(n_clusters) + '_clusters.tsv')
224 tmp.append(priorityAND(l[0])) 377
225 else: 378
226 tmp.append(l[0]) 379
227 l = l[1:] 380
228 elif l[0] == 'or': 381 ############################# main ###########################################
229 tmp.append(l[0]) 382
230 flag = False 383
231 l = l[1:] 384 def main():
232 elif l[1] == 'or': 385 if not os.path.exists('clustering'):
233 if isinstance(l[0], list): 386 os.makedirs('clustering')
234 tmp.append(priorityAND(l[0])) 387
235 else:
236 tmp.append(l[0])
237 tmp.append(l[1])
238 flag = False
239 l = l[2:]
240 elif l[1] == 'and':
241 tmpAnd = []
242 if isinstance(l[0], list):
243 tmpAnd.append(priorityAND(l[0]))
244 else:
245 tmpAnd.append(l[0])
246 tmpAnd.append(l[1])
247 if isinstance(l[2], list):
248 tmpAnd.append(priorityAND(l[2]))
249 else:
250 tmpAnd.append(l[2])
251 l = l[3:]
252 while l:
253 if l[0] == 'and':
254 tmpAnd.append(l[0])
255 if isinstance(l[1], list):
256 tmpAnd.append(priorityAND(l[1]))
257 else:
258 tmpAnd.append(l[1])
259 l = l[2:]
260 elif l[0] == 'or':
261 flag = False
262 break
263 if flag == True: #se ci sono solo AND nella lista
264 tmp.extend(tmpAnd)
265 elif flag == False:
266 tmp.append(tmpAnd)
267 return tmp
268
269 def checkRule(l):
270 if len(l) == 1:
271 if isinstance(l[0], list):
272 if checkRule(l[0]) is False:
273 return False
274 elif len(l) > 2:
275 if checkRule2(l) is False:
276 return False
277 else:
278 return False
279 return True
280
281 def checkRule2(l):
282 while l:
283 if len(l) == 1:
284 return False
285 elif isinstance(l[0], list) and l[1] in ['and', 'or']:
286 if checkRule(l[0]) is False:
287 return False
288 if isinstance(l[2], list):
289 if checkRule(l[2]) is False:
290 return False
291 l = l[3:]
292 elif l[1] in ['and', 'or']:
293 if isinstance(l[2], list):
294 if checkRule(l[2]) is False:
295 return False
296 l = l[3:]
297 elif l[0] in ['and', 'or']:
298 if isinstance(l[1], list):
299 if checkRule(l[1]) is False:
300 return False
301 l = l[2:]
302 else:
303 return False
304 return True
305
306 def do_rules(rules):
307 split_rules = []
308 err_rules = []
309 tmp_gene_in_rule = []
310 for i in range(len(rules)):
311 tmp = list(rules[i])
312 if tmp:
313 tmp, tmp_genes = check_and_doWord(tmp)
314 tmp_gene_in_rule.extend(tmp_genes)
315 if tmp is False:
316 split_rules.append([])
317 err_rules.append(rules[i])
318 else:
319 tmp = brackets_to_list(tmp)
320 if checkRule(tmp):
321 split_rules.append(priorityAND(tmp))
322 else:
323 split_rules.append([])
324 err_rules.append(rules[i])
325 else:
326 split_rules.append([])
327 if err_rules:
328 warning('Warning: wrong format rule in ' + str(err_rules) + '\n')
329 return (split_rules, list(set(tmp_gene_in_rule)))
330
331 def make_recon(data):
332 try:
333 import cobra as cb
334 import warnings
335 with warnings.catch_warnings():
336 warnings.simplefilter('ignore')
337 recon = cb.io.read_sbml_model(data)
338 react = recon.reactions
339 rules = [react[i].gene_reaction_rule for i in range(len(react))]
340 ids = [react[i].id for i in range(len(react))]
341 except cb.io.sbml3.CobraSBMLError:
342 try:
343 data = (pd.read_csv(data, sep = '\t', dtype = str)).fillna('')
344 if len(data.columns) < 2:
345 sys.exit('Execution aborted: wrong format of ' +
346 'custom GPR rules\n')
347 if not len(data.columns) == 2:
348 warning('WARNING: more than 2 columns in custom GPR rules.\n' +
349 'Extra columns have been disregarded\n')
350 ids = list(data.iloc[:, 0])
351 rules = list(data.iloc[:, 1])
352 except pd.errors.EmptyDataError:
353 sys.exit('Execution aborted: wrong format of custom GPR rules\n')
354 except pd.errors.ParserError:
355 sys.exit('Execution aborted: wrong format of custom GPR rules\n')
356 split_rules, tmp_genes = do_rules(rules)
357 gene_in_rule = {}
358 for i in tmp_genes:
359 gene_in_rule[i] = 'ok'
360 return (ids, split_rules, gene_in_rule)
361
362 ############################ resolve_methods ##################################
363
364 def replace_gene_value(l, d):
365 tmp = []
366 err = []
367 while l:
368 if isinstance(l[0], list):
369 tmp_rules, tmp_err = replace_gene_value(l[0], d)
370 tmp.append(tmp_rules)
371 err.extend(tmp_err)
372 else:
373 value = replace_gene(l[0],d)
374 tmp.append(value)
375 if value == None:
376 err.append(l[0])
377 l = l[1:]
378 return (tmp, err)
379
380 def replace_gene(l, d):
381 if l =='and' or l == 'or':
382 return l
383 else:
384 value = d.get(l, None)
385 if not(value == None or isinstance(value, (int, float))):
386 sys.exit('Execution aborted: ' + value + ' value not valid\n')
387 return value
388
389 def compute(val1, op, val2, cn):
390 if val1 != None and val2 != None:
391 if op == 'and':
392 return min(val1, val2)
393 else:
394 return val1 + val2
395 elif op == 'and':
396 if cn is True:
397 if val1 != None:
398 return val1
399 elif val2 != None:
400 return val2
401 else:
402 return None
403 else:
404 return None
405 else:
406 if val1 != None:
407 return val1
408 elif val2 != None:
409 return val2
410 else:
411 return None
412
413 def control(ris, l, cn):
414 if len(l) == 1:
415 if isinstance(l[0], (float, int)) or l[0] == None:
416 return l[0]
417 elif isinstance(l[0], list):
418 return control(None, l[0], cn)
419 else:
420 return False
421 elif len(l) > 2:
422 return control_list(ris, l, cn)
423 else:
424 return False
425
426 def control_list(ris, l, cn):
427 while l:
428 if len(l) == 1:
429 return False
430 elif (isinstance(l[0], (float, int)) or
431 l[0] == None) and l[1] in ['and', 'or']:
432 if isinstance(l[2], (float, int)) or l[2] == None:
433 ris = compute(l[0], l[1], l[2], cn)
434 elif isinstance(l[2], list):
435 tmp = control(None, l[2], cn)
436 if tmp is False:
437 return False
438 else:
439 ris = compute(l[0], l[1], tmp, cn)
440 else:
441 return False
442 l = l[3:]
443 elif l[0] in ['and', 'or']:
444 if isinstance(l[1], (float, int)) or l[1] == None:
445 ris = compute(ris, l[0], l[1], cn)
446 elif isinstance(l[1], list):
447 tmp = control(None,l[1], cn)
448 if tmp is False:
449 return False
450 else:
451 ris = compute(ris, l[0], tmp, cn)
452 else:
453 return False
454 l = l[2:]
455 elif isinstance(l[0], list) and l[1] in ['and', 'or']:
456 if isinstance(l[2], (float, int)) or l[2] == None:
457 tmp = control(None, l[0], cn)
458 if tmp is False:
459 return False
460 else:
461 ris = compute(tmp, l[1], l[2], cn)
462 elif isinstance(l[2], list):
463 tmp = control(None, l[0], cn)
464 tmp2 = control(None, l[2], cn)
465 if tmp is False or tmp2 is False:
466 return False
467 else:
468 ris = compute(tmp, l[1], tmp2, cn)
469 else:
470 return False
471 l = l[3:]
472 else:
473 return False
474 return ris
475
476 ############################ gene #############################################
477
478 def data_gene(gene, type_gene, name, gene_custom):
479 args = process_args(sys.argv) 388 args = process_args(sys.argv)
480 for i in range(len(gene)): 389
481 tmp = gene.iloc[i, 0] 390 #Data read
482 if tmp.startswith(' ') or tmp.endswith(' '): 391
483 gene.iloc[i, 0] = (tmp.lstrip()).rstrip() 392 X = read_dataset(args.input)
484 gene_dup = [item for item, count in 393 X = pd.DataFrame.to_dict(X, orient='list')
485 collections.Counter(gene[gene.columns[0]]).items() if count > 1] 394 X = rewrite_input(X)
486 pat_dup = [item for item, count in 395 X = pd.DataFrame.from_dict(X, orient = 'index')
487 collections.Counter(list(gene.columns)).items() if count > 1] 396
488 if gene_dup: 397 for i in X.columns:
489 if gene_custom == None: 398 tmp = X[i][0]
490 if args.rules_selector == 'HMRcore':
491 gene_in_rule = pk.load(open(args.tool_dir +
492 '/local/HMRcore_genes.p', 'rb'))
493 elif args.rules_selector == 'Recon':
494 gene_in_rule = pk.load(open(args.tool_dir +
495 '/local/Recon_genes.p', 'rb'))
496 gene_in_rule = gene_in_rule.get(type_gene)
497 else:
498 gene_in_rule = gene_custom
499 tmp = []
500 for i in gene_dup:
501 if gene_in_rule.get(i) == 'ok':
502 tmp.append(i)
503 if tmp:
504 sys.exit('Execution aborted because gene ID '
505 + str(tmp) + ' in ' + name + ' is duplicated\n')
506 if pat_dup:
507 sys.exit('Execution aborted: duplicated label\n'
508 + str(pat_dup) + 'in ' + name + '\n')
509 return (gene.set_index(gene.columns[0])).to_dict()
510
511 ############################ resolve ##########################################
512
513 def resolve(genes, rules, ids, resolve_none, name):
514 resolve_rules = {}
515 not_found = []
516 flag = False
517 for key, value in genes.items():
518 tmp_resolve = []
519 for i in range(len(rules)):
520 tmp = rules[i]
521 if tmp:
522 tmp, err = replace_gene_value(tmp, value)
523 if err:
524 not_found.extend(err)
525 ris = control(None, tmp, resolve_none)
526 if ris is False or ris == None:
527 tmp_resolve.append(None)
528 else:
529 tmp_resolve.append(ris)
530 flag = True
531 else:
532 tmp_resolve.append(None)
533 resolve_rules[key] = tmp_resolve
534 if flag is False:
535 sys.exit('Execution aborted: no computable score' +
536 ' (due to missing gene values) for class '
537 + name + ', the class has been disregarded\n')
538 return (resolve_rules, list(set(not_found)))
539
540 ################################# clustering ##################################
541
542 def f_cluster(resolve_rules):
543 os.makedirs('cluster_out')
544 args = process_args(sys.argv)
545 k_min = args.k_min
546 k_max = args.k_max
547 if k_min > k_max:
548 warning('k range boundaries inverted.\n')
549 tmp = k_min
550 k_min = k_max
551 k_max = tmp
552 else:
553 warning('k range correct.\n')
554 cluster_data = pd.DataFrame.from_dict(resolve_rules, orient = 'index')
555 for i in cluster_data.columns:
556 tmp = cluster_data[i][0]
557 if tmp == None: 399 if tmp == None:
558 cluster_data = cluster_data.drop(columns=[i]) 400 X = X.drop(columns=[i])
559 distorsion = [] 401
560 for i in range(k_min, k_max+1): 402 X = pd.DataFrame.to_numpy(X)
561 tmp_kmeans = KMeans(n_clusters = i, 403
562 n_init = 100, 404
563 max_iter = 300, 405 if args.cluster_type == 'kmeans':
564 random_state = 0).fit(cluster_data) 406 kmeans(args.k_min, args.k_max, X, args.elbow, args.silhouette, args.davies)
565 distorsion.append(tmp_kmeans.inertia_) 407
566 predict = tmp_kmeans.predict(cluster_data) 408 if args.cluster_type == 'dbscan':
567 predict = [x+1 for x in predict] 409 dbscan(X, args.eps, args.min_samples)
568 classe = (pd.DataFrame(list(zip(cluster_data.index, predict)))).astype(str) 410
569 dest = 'cluster_out/K=' + str(i) + '_' + args.name+'.tsv' 411 if args.cluster_type == 'hierarchy':
570 classe.to_csv(dest, sep = '\t', index = False, 412 hierachical_agglomerative(X, args.k_min, args.k_max)
571 header = ['Patient_ID', 'Class']) 413
572 plt.figure(0) 414 ##############################################################################
573 plt.plot(range(k_min, k_max+1), distorsion, marker = 'o')
574 plt.xlabel('Number of cluster')
575 plt.ylabel('Distorsion')
576 plt.savefig(args.elbow, dpi = 240, format = 'pdf')
577 if args.cond_hier == 'yes':
578 import scipy.cluster.hierarchy as hier
579 lin = hier.linkage(cluster_data, args.linkage)
580 plt.figure(1)
581 plt.figure(figsize=(10, 5))
582 hier.dendrogram(lin, leaf_font_size = 2, labels = cluster_data.index)
583 plt.savefig(args.dendro, dpi = 480, format = 'pdf')
584 return None
585
586 ################################# main ########################################
587
588 def main():
589 args = process_args(sys.argv)
590 if args.rules_selector == 'HMRcore':
591 recon = pk.load(open(args.tool_dir + '/local/HMRcore_rules.p', 'rb'))
592 elif args.rules_selector == 'Recon':
593 recon = pk.load(open(args.tool_dir + '/local/Recon_rules.p', 'rb'))
594 elif args.rules_selector == 'Custom':
595 ids, rules, gene_in_rule = make_recon(args.custom)
596 resolve_none = check_bool(args.none)
597 dataset = read_dataset(args.data, args.name)
598 dataset.iloc[:, 0] = (dataset.iloc[:, 0]).astype(str)
599 type_gene = gene_type(dataset.iloc[0, 0], args.name)
600 if args.rules_selector != 'Custom':
601 genes = data_gene(dataset, type_gene, args.name, None)
602 ids, rules = load_id_rules(recon.get(type_gene))
603 elif args.rules_selector == 'Custom':
604 genes = data_gene(dataset, type_gene, args.name, gene_in_rule)
605 resolve_rules, err = resolve(genes, rules, ids, resolve_none, args.name)
606 if err:
607 warning('WARNING: gene\n' + str(err) + '\nnot found in class '
608 + args.name + ', the expression level for this gene ' +
609 'will be considered NaN\n')
610 f_cluster(resolve_rules)
611 warning('Execution succeeded')
612 return None
613
614 ###############################################################################
615 415
616 if __name__ == "__main__": 416 if __name__ == "__main__":
617 main() 417 main()