-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathctec_multiple_method_ensemble.py
More file actions
347 lines (276 loc) · 16.1 KB
/
ctec_multiple_method_ensemble.py
File metadata and controls
347 lines (276 loc) · 16.1 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
# -*- coding: utf-8 -*-
## import system file
from operator import is_not
import os
import scanpy as sc
import numpy as np
import time
from utils.algo_ctec_multiple import CTEC_OB_multiple, CTEC_DB_multiple
from sklearn.metrics.cluster import adjusted_rand_score
from sklearn.metrics import calinski_harabasz_score
from utils import calc_acc
import pandas as pd
import argparse
import gc
import warnings
warnings.filterwarnings("ignore")
from utils.algo_ctec import post_process_cluster_name_with_order
def evaluate_df(df_input,pred_name,GTname):
df = df_input[[pred_name,GTname]]
acc_results_value, acc_results_name= calc_acc.calc_all_acc_simple(df,GTname,pred_name,decimals=4)
return acc_results_value, acc_results_name
def evaluate(adata,pred_name,GTname):
df = adata.obs[[pred_name,GTname]]
acc_results_value, acc_results_name= calc_acc.calc_all_acc_simple(df,GTname,pred_name,decimals=4)
return acc_results_value, acc_results_name
if __name__ == '__main__':
WORK_PATH = '/ctec_work'
DATA_PATH = os.path.join(WORK_PATH,'base_method_result/multi_method')
SAVE_PATH = WORK_PATH
parser = argparse.ArgumentParser()
parser.add_argument('--exp_id', type=str, default='demo', required=False, help='experiment id by int')
args = parser.parse_args()
EXP = args.exp_id
exist_ground_truth_cell_type=1
save_path = os.path.join(SAVE_PATH,'Result_Ensemble_Multi_Method_Exp_'+str(EXP))
try:
os.mkdir(save_path)
except:
pass
data_list_fullname = ['macaque_bc', 'human_pbmc_GSE96583', 'mouse_cortex_SCP425', 'human_pancreas', 'paul15']
algo_name_raw = ['SC3','CIDR', 'Seurat', 'Kmeans', 'SIMLR']
for i in range(0,len(data_list_fullname)):
data_name = data_list_fullname[i]
algo_name = algo_name_raw[:] #deep copy list
print('\n\n','='*100)
print('data_name=',data_name,', EXP=',EXP)
#[1.1 read data]
adata = sc.read(os.path.join(DATA_PATH,data_list_fullname[i]+'_result.h5ad'))
GTname = 'celltype'
ADATA_RAW = adata.copy()
algo_name = algo_name_raw[:] #deep copy list
print('algo list ** BEFORE** filtering:\n', algo_name)
algo_score_list = []
for algo in algo_name:
ch_score = calinski_harabasz_score(adata.obsm['X_pca'], adata.obs[algo])
algo_score_list.append((algo, ch_score))
algo_score_list.sort(key=lambda x: x[1], reverse=False)
algo_name.remove(algo_score_list[0][0]) # remove the worse algo based on CH score
print('algo list ** AFTER ** filtering:\n', algo_name)
num_algo = len(algo_name_raw)
print('running CTEC_DB')
CV_THRESHOLD = 0.5
ITER_NUMBER = 10
t0=time.time()
adata_ctec_db = CTEC_DB_multiple(ADATA_RAW,algo_name,CV_THRESHOLD, ITER_NUMBER,PRINT_INFO=True)
time_ctec_db=time.time()-t0
adata_ctec_db = post_process_cluster_name_with_order(adata_ctec_db, 'cluster_ensemble')
adata_ctec_db.obs['cluster_ensemble'] = pd.Categorical( adata_ctec_db.obs['cluster_ensemble'].astype(str) )
ADATA_RAW.obs['ctec_db_labels'] = adata_ctec_db.obs['cluster_ensemble']
ADATA_RAW.obs.to_csv(os.path.join(save_path, '{}_processed_CTECDB_{}algo.csv'.format(data_name, num_algo)))
if exist_ground_truth_cell_type:
acc_ctec_db, acc_results_name = evaluate(ADATA_RAW,'ctec_db_labels',GTname)
print('\n\n____________acc: ctec_db_labels')
print(acc_results_name,acc_ctec_db)
result_acc_path = os.path.join(save_path,'result_acc=={}==_CTECDB_{}algo.csv'.format(data_name, num_algo))
with open(result_acc_path ,'a+') as f:
f.write('File,'+data_name+'\n')
f.write('ITER_NUMBER,'+str(ITER_NUMBER)+'\n')
f.write('Total_time,'+str(time_ctec_db)+'\n')
f.write('Acc_Name,'+calc_acc.list2str(acc_results_name)+','+'resolution'+','+'time_consuming'+'\n')
f.write('Acc_ctec_db,'+calc_acc.list2str(acc_ctec_db)+','+'NA ,'+str(time_ctec_db)+'\n')
f.write('\n')
else:
pass
print('\n running CTEC_OB')
ITER_NUMBER = 10
t0=time.time()
adata_ctec_ob = CTEC_OB_multiple(ADATA_RAW,algo_name,ITER_NUMBER,PRINT_INFO=True)
time_ctec_ob=time.time()-t0
adata_ctec_ob = post_process_cluster_name_with_order(adata_ctec_ob, 'cluster_ensemble')
adata_ctec_ob.obs['cluster_ensemble'] = pd.Categorical( adata_ctec_ob.obs['cluster_ensemble'].astype(str) )
ADATA_RAW.obs['ctec_ob_labels'] = adata_ctec_ob.obs['cluster_ensemble']
ADATA_RAW.obs.to_csv((os.path.join(save_path, '{}_processed_CTECOB_{}algo.csv'.format(data_name, num_algo))))
if exist_ground_truth_cell_type:
acc_ctec_ob, acc_results_name = evaluate(ADATA_RAW,'ctec_ob_labels',GTname)
print('\n\n____________acc: ctec_ob_labels')
print(acc_results_name,acc_ctec_ob)
result_acc_path = os.path.join(save_path,'result_acc=={}==_CTECOB_{}algo.csv'.format(data_name, num_algo))
with open(result_acc_path ,'a+') as f:
f.write('File,'+data_name+'\n')
f.write('ITER_NUMBER,'+str(ITER_NUMBER)+'\n')
f.write('Total_time,'+str(time_ctec_ob)+'\n')
f.write('Acc_Name,'+calc_acc.list2str(acc_results_name)+','+'resolution'+','+'time_consuming'+'\n')
f.write('Acc_ctec_ob,'+calc_acc.list2str(acc_ctec_ob)+','+'NA ,'+str(time_ctec_ob)+'\n')
f.write('\n')
else:
pass
# -------------------------------------------------------------------------------
## acc summary
adata_raw_obs_df = pd.DataFrame(ADATA_RAW.obs)
ground_truth_df = adata_raw_obs_df[GTname].astype(str)
algo_ctec_ob_df = adata_raw_obs_df['ctec_ob_labels'].astype(str)
algo_ctec_db_df = adata_raw_obs_df['ctec_db_labels'].astype(str)
algo1_df = adata_raw_obs_df[algo_name_raw[0]].astype(str)
algo2_df = adata_raw_obs_df[algo_name_raw[1]].astype(str)
algo3_df = adata_raw_obs_df[algo_name_raw[2]].astype(str)
algo4_df = adata_raw_obs_df[algo_name_raw[3]].astype(str)
algo5_df = adata_raw_obs_df[algo_name_raw[4]].astype(str)
SAFE_df = adata_raw_obs_df['SAFE'].astype(str)
SAME_df = adata_raw_obs_df['SAME'].astype(str)
print('\n---- Summary ----')
DF_FOR_ACC1 = pd.concat([algo1_df, ground_truth_df], axis=1, join="inner")
acc_algo1, acc_results_name = evaluate_df(DF_FOR_ACC1,algo_name_raw[0],GTname)
print(acc_results_name,acc_algo1,algo_name_raw[0])
DF_FOR_ACC2 = pd.concat([algo2_df, ground_truth_df], axis=1, join="inner")
acc_algo2, acc_results_name = evaluate_df(DF_FOR_ACC2,algo_name_raw[1],GTname)
print(acc_results_name,acc_algo2,algo_name_raw[1])
DF_FOR_ACC3 = pd.concat([algo3_df, ground_truth_df], axis=1, join="inner")
acc_algo3, acc_results_name = evaluate_df(DF_FOR_ACC3,algo_name_raw[2],GTname)
print(acc_results_name,acc_algo3,algo_name_raw[2])
DF_FOR_ACC4 = pd.concat([algo4_df, ground_truth_df], axis=1, join="inner")
acc_algo4, acc_results_name = evaluate_df(DF_FOR_ACC4,algo_name_raw[3],GTname)
print(acc_results_name,acc_algo4,algo_name_raw[3])
DF_FOR_ACC5 = pd.concat([algo5_df, ground_truth_df], axis=1, join="inner")
acc_algo5, acc_results_name = evaluate_df(DF_FOR_ACC5,algo_name_raw[4],GTname)
print(acc_results_name,acc_algo5,algo_name_raw[4])
DF_FOR_SAFE = pd.concat([SAFE_df, ground_truth_df], axis=1, join="inner")
acc_SAFE, acc_results_name = evaluate_df(DF_FOR_SAFE,'SAFE',GTname)
print(acc_results_name,acc_SAFE,'SAFE')
DF_FOR_SAME = pd.concat([SAME_df, ground_truth_df], axis=1, join="inner")
acc_SAME, acc_results_name = evaluate_df(DF_FOR_SAME,'SAME',GTname)
print(acc_results_name,acc_SAME,'SAME')
DF_FOR_ACC_ctec_ob = pd.concat([algo_ctec_ob_df, ground_truth_df], axis=1, join="inner")
acc_ctec_ob, acc_results_name = evaluate_df(DF_FOR_ACC_ctec_ob,'ctec_ob_labels',GTname)
print(acc_results_name,acc_ctec_ob,'CTEC_OB')
DF_FOR_ACC_ctec_db = pd.concat([algo_ctec_db_df, ground_truth_df], axis=1, join="inner")
acc_ctec_db, acc_results_name = evaluate_df(DF_FOR_ACC_ctec_db,'ctec_db_labels',GTname)
print(acc_results_name,acc_ctec_db,'CTEC_DB')
result_acc_path = os.path.join(save_path,'result_acc=='+data_name+'==_CTEC_multiple.csv')
with open(result_acc_path ,'a+') as f:
f.write('File,'+data_name+'\n')
f.write('Acc_Name,'+calc_acc.list2str(acc_results_name)+','+'time_consuming'+'\n')
f.write('Acc_'+algo_name_raw[0]+','+calc_acc.list2str(acc_algo1)+','+str('')+'\n')
f.write('Acc_'+algo_name_raw[1]+','+calc_acc.list2str(acc_algo2)+','+str('')+'\n')
f.write('Acc_'+algo_name_raw[2]+','+calc_acc.list2str(acc_algo3)+','+str('')+'\n')
f.write('Acc_'+algo_name_raw[3]+','+calc_acc.list2str(acc_algo4)+','+str('')+'\n')
f.write('Acc_'+algo_name_raw[4]+','+calc_acc.list2str(acc_algo5)+','+str('')+'\n')
f.write('Acc_'+'SAFE'+','+calc_acc.list2str(acc_SAFE)+','+str('')+'\n')
f.write('Acc_'+'SAME'+','+calc_acc.list2str(acc_SAME)+','+str('')+'\n')
f.write('Acc_CTEC_OB,'+calc_acc.list2str(acc_ctec_ob)+','+str(time_ctec_ob)+'\n')
f.write('Acc_CTEC_DB,'+calc_acc.list2str(acc_ctec_db)+','+str(time_ctec_db)+'\n')
f.write('\n')
del ADATA_RAW
del adata
gc.collect()
print('data_name_NUM=',i,', EXP=',EXP)
print('='*100)
def draw_umap(i):
num_algo = len(algo_name_raw)
data_name = data_list_fullname[i]
print('\n\n','='*100)
print('data_name=',data_name,', EXP=',EXP)
#[1.1 read data]
adata = sc.read(os.path.join(DATA_PATH,data_list_fullname[i]+'_res.h5ad'))
GTname = 'celltype'
from matplotlib import pyplot as plt
import matplotlib
matplotlib.use('Agg')
from utils.umap_util import get_color_dict
save_path_umap = os.path.join(save_path,'umap_result')
try:
os.mkdir(save_path_umap)
except:
pass
FIG_DPI = 300
t0=time.time()
# data_name, data_ext,GTname,class_nums,BATCH_KEY,N_HIGH_VAR,res_leiden,res_louvain = loaddata_default(DATASET_NUM)
print('\n\n','='*100)
print('Step 7: calc UMAP')
print( 'EXP=',EXP)
print('data_name = ',data_name)
print('='*100)
t_start_dataset = time.time()
## load dataframe CTEC-DB
PATH_ctec_db_labels_df = os.path.join(save_path, '{}_processed_CTECDB_{}algo.csv'.format(data_name, num_algo))
ctec_db_labels_df = pd.read_csv(PATH_ctec_db_labels_df,index_col=0)
## load dataframe CTEC-OB
PATH_ctec_ob_labels_df = os.path.join(save_path, '{}_processed_CTECOB_{}algo.csv'.format(data_name, num_algo))
ctec_ob_labels_df = pd.read_csv(PATH_ctec_ob_labels_df,index_col=0)
plot_obsname = [GTname,algo_name_raw[0],algo_name_raw[1],algo_name_raw[2],algo_name_raw[3],algo_name_raw[4],'SAFE','SAME','ctec_db_labels','ctec_ob_labels']
plot_showname = ['Reference standard annotation','SC3','CIDR', 'Seurat', 'Kmeans', 'SIMLR','SAFE','SAME','CTEC-DB','CTEC-OB']
save_name = data_name
SAVE_PDF = 1
adata.obs['ctec_db_labels'] = pd.Categorical(ctec_db_labels_df['ctec_db_labels'])
adata.obs['ctec_ob_labels'] = pd.Categorical(ctec_ob_labels_df['ctec_ob_labels'])
sc.tl.umap(adata)
sc.settings.set_figure_params(dpi_save=FIG_DPI, figsize=(8, 6), facecolor='white',) # low dpi (dots per inch) yields small inline figures
for i in range(0,len(plot_obsname)):
try:
adata.obs[plot_showname[i]] = adata.obs[plot_obsname[i]].astype(str)
except:
adata.obs[plot_showname[i]] = adata.obs[plot_obsname[i]].astype(str)
# tmp1 = list(set(list(adata.obs[plot_obsname[i]])))
# tmp2 = list(set(list(adata.obs[plot_showname[i]])))
# try:
# tmp1.sort()
# tmp2.sort()
# except:
# pass
# print('tmp1 = ',tmp1)
# print('tmp2 = ',tmp2)
## Plot UMAP with same color for different algo
## for Ground Truth result
COLORLIST33 = ['#FFFF00','#1CE6FF','#FF34FF','#FF4A46','#008941','#006FA6','#A30059','#FFDBE5','#7A4900','#0000A6','#63FFAC','#B79762','#004D43','#8FB0FF','#997D87','#5A0007','#809693','#FEFFE6','#1B4400','#4FC601','#3B5DFF','#4A3B53','#FF2F80','#61615A','#BA0900','#6B7900','#00C2A0','#FFAA92','#FF90C9','#B903AA','#D16100','#DDEFFF','#000035']
colorlist = COLORLIST33.copy()
group_name = 'Reference standard annotation'
gt_namelist = list(set(adata.obs[group_name]))
gt_namelist.sort()
colordict_gt = dict()
for i in range(0,len(gt_namelist)):
colordict_gt[gt_namelist[i]] = colorlist[i]
with plt.rc_context(): # Use this to set figure params like size and dpi
sc.pl.umap(adata,
color=group_name,
palette=colordict_gt,
legend_loc = 'on data',
legend_fontsize = 'xx-small',
legend_fontoutline=2)
plt.savefig(os.path.join(save_path_umap, data_name+'_'+group_name+'_UMAP_2Algo_OnData.pdf'),bbox_inches='tight',dpi=FIG_DPI)
plt.savefig(os.path.join(save_path_umap, data_name+'_'+group_name+'_UMAP_2Algo_OnData.png'),bbox_inches='tight',dpi=FIG_DPI)
with plt.rc_context(): # Use this to set figure params like size and dpi
sc.pl.umap(adata,
color=group_name,
palette=colordict_gt,
legend_loc = 'right margin',
legend_fontsize = 'xx-small',
legend_fontoutline=2)
plt.savefig(os.path.join(save_path_umap, data_name+'_'+group_name+'_UMAP_2Algo_Right.pdf'),bbox_inches='tight',dpi=FIG_DPI)
plt.savefig(os.path.join(save_path_umap, data_name+'_'+group_name+'_UMAP_2Algo_Right.png'),bbox_inches='tight',dpi=FIG_DPI)
## for different algorithm result
for j in range(1,len(plot_showname)):
group_name=plot_showname[j]
name_gt = 'Reference standard annotation'
##################################################################
colordict_gp = get_color_dict(adata,name_gt,group_name,colordict_gt,COLORLIST33.copy())
##################################################################
with plt.rc_context(): # Use this to set figure params like size and dpi
sc.pl.umap(adata,
color=group_name,
palette=colordict_gp,
legend_loc = 'right margin',
legend_fontsize = 'xx-small',
legend_fontoutline=2)
plt.savefig(os.path.join(save_path_umap, data_name+'_'+group_name+'_UMAP_2Algo_Right.pdf'),bbox_inches='tight',dpi=FIG_DPI)
plt.savefig(os.path.join(save_path_umap, data_name+'_'+group_name+'_UMAP_2Algo_Right.png'),bbox_inches='tight',dpi=FIG_DPI)
with plt.rc_context(): # Use this to set figure params like size and dpi
sc.pl.umap(adata,
color=group_name,
palette=colordict_gp,
legend_loc = 'on data',
legend_fontsize = 'xx-small',
legend_fontoutline=2)
plt.savefig(os.path.join(save_path_umap, data_name+'_'+group_name+'_UMAP_2Algo_OnData.pdf'),bbox_inches='tight',dpi=FIG_DPI)
plt.savefig(os.path.join(save_path_umap, data_name+'_'+group_name+'_UMAP_2Algo_OnData.png'),bbox_inches='tight',dpi=FIG_DPI)
print('umap image Saved:'+ os.path.join(save_path_umap, data_name+'_'+group_name+'_UMAP_OnData......'))
print('total calc time = ',time.time()-t0)