-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathTissueSegmentation.py
1133 lines (1009 loc) · 39.9 KB
/
TissueSegmentation.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
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
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
import os, sys
os.environ['OPENCV_IO_MAX_IMAGE_PIXELS'] = "1000000000000"
import cv2
import tifffile
import pandas as pd
import numpy as np
import matplotlib
matplotlib.use('AGG')
import matplotlib.pyplot as plt
from scipy import ndimage
from skimage import filters
import glob
import time
import gzip
import csv
from shutil import copyfile
import scipy.signal as signal
import seaborn as sns
import gefpy
from gefpy.bgef_writer_cy import generate_bgef
from gefpy import plot as gefplot
import h5py
from itertools import groupby
from ImageTools import CreatImg, readDNB
from ImagePyramid import createPyramid
import tissue_cut.tissueCut_utils.tissue_seg_pipeline as tissue_seg_pipeline
import gc
import glog
class tissueSegmentation():
def __init__(self, infile, outpath, binSize, snId, develop, omics):
self.t0 = time.time()
self.infile = infile
self.snId = snId
# check input file format
try:
f = h5py.File(self.infile, 'r')
f.close()
except:
# txt format
# self.typeColumn = {"geneID": 'category', "x": np.uint32, "y": np.uint32, "values": np.uint32, "UMICount": np.uint32, "MIDCount": np.uint32}
# self.genedf = pd.read_csv(infile, sep='\t', dtype=self.typeColumn, quoting=csv.QUOTE_NONE, comment="#")
# self.resolution = [0]
# self.version = [2]
raise ("invalid gef format")
# else:
# self.loadGef()
t1 = time.time()
print("Loading data done. time used: {:.3f}".format(t1 - self.t0))
# self.genedf['x'] = self.genedf['x'] - self.genedf['x'].min()
# self.genedf['y'] = self.genedf['y'] - self.genedf['y'].min()
# self.x1, self.x2 = self.genedf['x'].min(), self.genedf['x'].max()
# self.y1, self.y2 = self.genedf['y'].min(), self.genedf['y'].max()
# self.ori_shape = (self.y2 - self.y1 + 1, self.x2 - self.x1 + 1)
# self.total_umi = self.genedf['MIDCount'].sum()
self.total_umi=np.int64(0)
self.outpath = outpath
self.binSize = binSize
self.figpath = os.path.join(self.outpath, "tissue_fig")
self.stainUsed = False
os.makedirs(self.figpath, exist_ok=True)
self.develop = develop
self.omics = omics
def loadGef(self):
f = h5py.File(self.infile, 'r')
self.rawGefMinX = f['/geneExp/bin1/expression'].attrs['minX'][0]
self.rawGefMinY = f['/geneExp/bin1/expression'].attrs['minY'][0]
exp = f['/geneExp/bin1/expression']
gene = f['/geneExp/bin1/gene']
self.genedf = pd.DataFrame(exp[0:])
self.genedf['x'] += self.rawGefMinX
self.genedf['y'] += self.rawGefMinY
genes = []
for t in gene[0:]:
genes.extend([t[0].decode('utf-8')] * t[2])
self.genedf['geneID'] = np.array(genes)
self.genedf.rename(columns={'count': 'MIDCount'}, inplace=True)
self.genedf = self.genedf[['geneID', 'x', 'y', 'MIDCount']]
self.resolution = exp.attrs['resolution'][0]
self.version = f.attrs['version'][0]
self.tissueOffsetX = 0
self.tissueOffsetY = 0
f.close()
# Parse offset from SN.gef, for correction of tissue.gef
def parseGefOffset(self, filename):
f = h5py.File(self.infile, 'r')
self.snOffsetX = f['/geneExp/bin1/expression'].attrs['minX'][0]
self.snOffsetY = f['/geneExp/bin1/expression'].attrs['minY'][0]
self.resolution = f['/geneExp/bin1/expression'].attrs['resolution']
self.version = f.attrs['version']
f.close()
def process(self, stainfile, dnbfile, flip_code, multi, amp_shape, amp_factor, min_mean, platform, struct_kernel,
low_thresh, high_thresh):
t1 = time.time()
# maskTif = os.path.join(self.outpath, "bin1_mask.tif")
# mask=tifffile.imread(maskTif)
# glog.info('start dumpResult')
# conArea=949038458
# self.Dumpresult(mask, conArea)
# # mergedf, coor, num_dnb = self.Dumpresult(mask)
# glog.info('dumpResult done')
# return
ImgPath = os.path.join(self.outpath, "bin1_img.tif")
tissueCut = tissue_seg_pipeline.tissueCut(ImgPath, self.outpath, 0, 1)
if stainfile is None or not os.path.exists(stainfile):
glog.info("start drawBin100imgAndcompleteGef")
self.drawBin100imgAndcompleteGef()
glog.info("end drawBin100imgAndcompleteGef")
bin1Img=self.CreatImgFromFile(1,normalize=False)
glog.info("end drawBin1imgAndcompleteGef")
# print("Detecting contours.. ")
# print("start createImg")
# binImg = CreatImg(self.genedf, self.binSize)
# cellMask = None
# offsetX, offsetY, binImg = self.DnbMatting(binImg)
# print("end createImg")
# np.savetxt("bin100.txt", binImg, fmt='%d')
# print("end save")
# return 0
# Img = CreatImg(self.genedf, 1, normalize=False)
# Img = CreatImg(self.genedf, binSizeNoStain, normalize=False)
bin1Shape=bin1Img.shape
cv2.imwrite(ImgPath, bin1Img)
del bin1Img
gc.collect()
glog.info("start to tissueCut")
tissueCut.tissue_seg()
gc.collect()
glog.info("tissueCut completed")
# print (self.ori_shape)
# print (self.x1, self.x2, self.y1, self.y2)
mask = tissueCut.mask[0]
# print (mask.shape)
# np.savetxt("bin{0}_mask.txt".format(binSizeNoStain),mask)
maskTif=os.path.join(self.outpath, "bin1_mask.tif")
cv2.imwrite(maskTif,mask)
# return 0
conArea=np.int64(0)
# for x in range(mask.shape[0]):
# for y in range(mask.shape[1]):
# if mask[x][y]>0:
# conArea+=1
# conArea = len(np.where(mask > 0)[0])
conArea=np.count_nonzero(mask)
glog.info("conArea:\t"+str(conArea))
if (conArea == 0):
mask = np.ones(bin1Shape, np.uint8)
conArea = bin1Shape[0] * bin1Shape[1]
# cnts = self.FindContours(binImg, amp_shape, amp_factor, min_mean, platform, struct_kernel, low_thresh, high_thresh)
# t2 = time.time()
# print("Time used: {:.3f}".format(t2 - t1))
# if len(cnts) == 0:
# mask = np.ones(self.ori_shape, np.uint8)
# conArea = self.ori_shape[0] * self.ori_shape[1]
# t3 = time.time()
# else:
# print("Drawing contours...")
# cnt = sorted(cnts, key=cv2.contourArea, reverse=True)[0]
# conArea, mask = self.CreateMask(cnt, self.binSize)
# tmp = np.stack((binImg,) * 3, axis=2)
# cv2.drawContours(tmp, [cnt], -1, (0, 0, 255), 5)
# cv2.imwrite(os.path.join(self.outpath, "contour_image.tif"), tmp)
# t3 = time.time()
# print("Drawing contours used: {:.3f}".format(t3 - t2))
else:
t2 = time.time()
# filterGenefile = os.path.join(self.outpath, "{0}.gef".format(self.snId))
# self.parseGefOffset(filterGenefile)
self.stainUsed = True
conArea, mask, cellMask, offsetX, offsetY = self.stainSegmentation(stainfile, flip_code, multi)
# tmp = np.stack((binImg,) * 3, axis=2)
# cv2.drawContours(tmp, [cnt], -1, (0, 0, 255), 10)
# cv2.imwrite(os.path.join(outpath, "contour_image.tif"), tmp)
if(conArea==0):
print("Warning! mask is empty")
return
glog.info('start dumpResult')
self.Dumpresult(mask,conArea)
# mergedf, coor, num_dnb = self.Dumpresult(mask)
glog.info('dumpResult done')
# the rest steps will be replaced with c++ code, so some data in memory can be released
# del mergedf
del mask
del tissueCut
# del self.genedf
gc.collect()
glog.info("start stat")
# run c++ code
import ctypes
programDir = os.path.dirname(__file__)
if programDir == '':
programDir = './'
tsLibPath = os.path.join(programDir, 'ts.so')
tslib = ctypes.cdll.LoadLibrary(tsLibPath)
tslib.argtypes = [ctypes.POINTER(ctypes.c_char)]
if (self.omics == 'Proteomics'):
tissueGef = os.path.join(self.outpath, "{0}.protein.tissue.gef".format(self.snId))
else:
tissueGef = os.path.join(self.outpath, "{0}.tissue.gef".format(self.snId))
parameters = tissueGef
if dnbfile is None:
dnbfile='_VIRTUAL_FILE_PATH_'
parameters += "," + dnbfile
parameters += "," + str(self.tissueOffsetX)
parameters += "," + str(self.tissueOffsetY)
parameters += "," + str(self.snOffsetX)
parameters += "," + str(self.snOffsetY)
parameters += "," + str(conArea)
parameters += "," + str(self.total_umi)
parameters += "," + self.outpath
parameters += "," + str(self.stainUsed)
parameters += "," + str(self.develop)
parameters += ",200"
strParameters = bytes(parameters, 'utf-8')
print(parameters)
# string mergedf_File, string dnbFile, int offsetX, int offsetY, int coorOffsetX, int coorOffsetY, uint64_t conArea, uint64_t total_umi, string outpath, bool useStain, int maxbinSize;
tslib.tissueStat(strParameters)
glog.info("end stat")
# print("Dumping result used: {:.3f}".format(t5-t3))
# geneCount = mergedf['geneID'].nunique()
# print("Gene type: ", geneCount)
# umiCount = mergedf['UMICount'].sum()
# print("Total umi: ", umiCount)
# t6 = time.time()
# print("Count gene and umi used: {:.3f}".format(t6 - t5))
t5 = time.time()
# print(time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(time.time())), '\tstart readDNB and merge')
# if dnbfile != None:
# tmpdf = readDNB(dnbfile, offsetX, offsetY)
# dnbreads = tmpdf['reads'].sum()
# dnbdf = pd.merge(coor, tmpdf, how='inner')
# else:
# dnbdf = pd.DataFrame(columns=['x', 'y', 'reads'])
# dnbreads = 0
# t7 = time.time()
# print("Load dnb file used: {:.3f}".format(t7-t5))
# print(time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(time.time())), '\tstart TissueStat')
# self.TissueStat(mergedf, dnbdf, conArea, num_dnb, dnbreads)
# draw plot
for b in range(50, 200 + 1, 50):
dataFile = os.path.join(self.outpath, "bin{0}.midAndGeneCount.txt".format(b))
my_data = np.loadtxt(dataFile)
scapath = os.path.join(self.figpath, "scatter_{0}x{0}_MID_gene_counts.png".format(b if b != 0 else "cell"))
violinpath = os.path.join(self.figpath, "violin_{0}x{0}_MID_gene.png".format(b if b != 0 else "cell"))
statisticPath = os.path.join(self.figpath,
"statistic_{0}x{0}_MID_gene_DNB.png".format(b if b != 0 else "cell"))
plt.figure(figsize=(5, 5))
# sns.scatterplot(x=df['n_counts'], y=df['n_genes'], edgecolor="gray", color="gray")
plt.scatter(my_data[:, 0], my_data[:, 1], color="gray", edgecolors="gray", s=0.8)
plt.grid()
plt.xlabel("MID Count")
plt.ylabel("Gene Number")
plt.savefig(scapath, format="png", bbox_inches="tight")
plt.figure(figsize=(10, 6))
plt.subplot(121)
sns.violinplot(y=my_data[:, 0])
sns.stripplot(y=my_data[:, 0], jitter=0.4, color="black", size=0.8)
plt.ylabel("")
plt.title("MID Count")
plt.subplot(122)
sns.violinplot(y=my_data[:, 1])
sns.stripplot(y=my_data[:, 1], jitter=0.4, color="black", size=0.8)
plt.ylabel("")
plt.title("Gene Number")
plt.savefig(violinpath, format="png", bbox_inches="tight")
if self.develop:
tmpDevDf = pd.DataFrame()
tmpDevDf['MID Count'] = my_data[:, 0]
tmpDevDf['Gene Number'] = my_data[:, 1]
tmpDevDf['DNB Number'] = my_data[:, 2]
sns.set(font_scale=1.8)
sns.set_style("whitegrid")
g = sns.FacetGrid(pd.melt(tmpDevDf[['MID Count', 'Gene Number', 'DNB Number']]), col='variable',
hue='variable', sharex=False, sharey=False, height=8, palette='Set1')
g = (g.map(sns.distplot, "value", hist=False, rug=True))
plt.savefig(statisticPath)
sns.reset_defaults()
plt.close('all')
os.remove(dataFile)
glog.info("end draw pictures")
print(time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(time.time())), '\tTissueStat done')
t8 = time.time()
# print("Get stat result used: {:.3f}".format(t8 - t7))
print("Total segmentation time used: {:.2f}".format(t8 - self.t0))
def AmpImg(self, image, shape, ampfactor):
"""图像增强"""
print("Amplification starts..")
ampfilter = np.ones((shape, shape)) * ampfactor
amp_fig = ndimage.convolve(image, ampfilter, mode='constant')
print("Amplification done.")
return amp_fig
def CalGradient(self, binimg):
"""计算gradient"""
gradx = cv2.Sobel(binimg, ddepth=cv2.CV_32F, dx=1, dy=0, ksize=-1)
grady = cv2.Sobel(binimg, ddepth=cv2.CV_32F, dx=0, dy=1, ksize=-1)
# subtract the y-gradient from the x-gradient
gradient = cv2.subtract(gradx, grady)
gradient = cv2.convertScaleAbs(gradient)
return gradient
def Local_equalize(self, img, gridSize=5):
""" 局部直方图均衡 """
calhe = cv2.createCLAHE(clipLimit=5.0, tileGridSize=(gridSize, gridSize))
gray = calhe.apply(img)
return gray
def FindThreshold(self, blurred, low_thresh, high_thresh):
"""
根据表达量计算过滤的阈值
kernel: kernel size, default=10
low, high: the min and max pixel value for finding threshold
default low = 180, high = 255
"""
print("Threshold reveived: ", low_thresh, high_thresh)
# blurred = cv2.blur(gradient_img, (10, 10))
if low_thresh and high_thresh:
low = low_thresh
high = high_thresh
elif low_thresh and not high_thresh:
low = low_thresh
high = blurred.max()
else:
value = blurred[blurred > 0]
if len(value) == 0:
print("Failed to find threshold. exit")
sys.exit(1)
if not high_thresh:
high = 255
else:
high = high_thresh
hist = cv2.calcHist([value], [0], None, [256], [0, 255])
frq = np.concatenate(hist)
frq = signal.medfilt(frq, kernel_size=11)
peaks, prop = signal.find_peaks(frq, width=10, height=1.5 * frq.mean())
if len(peaks) == 0:
low = 220
else:
low = min(int(peaks[-1] - prop['widths'][-1] * 0.5), 250)
i = int(low)
lowpath = os.path.join(self.outpath, "frequency_plot.png")
plt.figure()
plt.plot(frq)
plt.plot(i, frq[i], "o")
plt.plot(peaks, frq[peaks], "x")
plt.savefig(lowpath)
print("threshold has been set as: ", low, high)
(_, thresh) = cv2.threshold(blurred, low, high, cv2.THRESH_BINARY)
return thresh
def FillHoles(self, thresh, struct_kernel):
"""去噪"""
kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (struct_kernel, struct_kernel))
closed = cv2.morphologyEx(thresh, cv2.MORPH_CLOSE, kernel)
open_kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (11, 11))
erodeimg = cv2.erode(closed, open_kernel, iterations=5)
dilateimg = cv2.dilate(erodeimg, open_kernel, iterations=10)
return dilateimg
def CreateMask(self, cnts, binSize):
"""生成mask"""
contours = []
for i in cnts:
contours.append([[i[0][0] * binSize, i[0][1] * binSize]])
contours = np.array(contours)
Area = cv2.contourArea(contours)
print("Contour area: ", Area)
#### Smooth mask
mask = np.zeros(self.ori_shape, np.uint8)
epsilon = 0.01 * cv2.arcLength(contours, True)
approx = cv2.approxPolyDP(contours, epsilon, True)
cv2.fillPoly(mask, [approx], (255, 0, 0))
mask = cv2.GaussianBlur(mask, (55, 55), 0)
# cnt_mask = np.zeros(self.ori_shape, np.uint8)
# cv2.drawContours(cnt_mask, [approx], -1, 255, 10)
cv2.imwrite(os.path.join(self.outpath, "approx_contour_mask.tiff"), mask)
return Area, mask
def stainSegmentation(self, stainfile, flip_code, multi):
"""
根据配准图片抠图
输入配准好的底图，
输出 对应可视化坐标 HE_x1_y1_height_width.png 的配准图，以及根据配准图生成的mask
"""
if not os.path.exists(stainfile):
print("Staining image not exist. ")
mask = np.ones(self.ori_shape, np.uint8)
conArea = self.ori_shape[0] * self.ori_shape[1]
return conArea, mask, None, 0, 0
try:
bbox = glob.glob(stainfile + "/*tissue_bbox.csv")[0]
except IOError:
sys.stderr.write("bbox file dose not exists, please check...")
sys.exit(1)
bdf = pd.read_csv(bbox, sep="\t")
cnt = bdf.loc(0)[0].values
# flip image verticality
# padding = 0
# x1, y1, x2, y2 = [cnt[0] - padding + self.x1, cnt[1] - padding + self.y1, cnt[2] + padding + self.x1,
# cnt[3] + padding + self.y1]
#
merPath = os.path.join(self.outpath, "{0}.gef".format(self.snId))
# merDf = self.genedf
# merDf = merDf.loc[(merDf['x'] >= x1) & (merDf['x'] < x2) & (merDf['y'] >= y1) & (merDf['y'] < y2)]
#
# merDf['x'] = merDf['x'] - x1
# merDf['y'] = merDf['y'] - y1
#
# self.genedf = merDf
# self.x1 = 0
# self.y1 = 0
# self.x2 = x2 - x1
# self.y2 = y2 - y1
# self.ori_shape = (self.y2 - self.y1 + 1, self.x2 - self.x1 + 1)
# print("x1: {0}\tx2: {1}\ty1: {2}\ty2: {3}".format(x1, x2, y1, y2))
#
# self.info = "#FileFormat=GEMv0.1\n#SortedBy=None\n#BinSize=1\n#StereoChip={0}\n#OffsetX={1}\n#OffsetY={2}\n".format(
# self.snId, x1, y1)
# with gzip.open(merPath, 'wt') as writer:
# self.info="#FileFormat=GEMv0.1\n#SortedBy=None\n#BinSize=1\n#StereoChip={0}\n#OffsetX={1}\n#OffsetY={2}\n".format(self.snId, x1, y1)
# writer.write(self.info)
# merDf[['geneID', 'x', 'y', 'MIDCount']].to_csv(writer, index=None, sep="\t")
# generate_bgef(self.infile, merPath,stromics=self.omics,region=[x1 - self.rawGefMinX, x2 - self.rawGefMinX, y1 - self.rawGefMinY, y2 - self.rawGefMinY])
self.parseGefOffset(merPath)
self.tissueOffsetX = self.snOffsetX
self.tissueOffsetY = self.snOffsetY
temppath = os.path.join(self.outpath, "dnb_merge")
os.makedirs(temppath, exist_ok=True)
gefplot.save_exp_heat_map(merPath, os.path.join(temppath, "bin200.png"))
# generate_bgef(self.infile, merPath+".multi.gef", bin_sizes=[1,10,20,50,100,200,500])
# merDf.to_csv(merPath, sep="\t", index=None)
# self.x1, self.x2, self.y1, self.y2 = self.genedf['x'].min(), self.genedf['x'].max(), self.genedf['y'].min(), self.genedf['y'].max()
# self.ori_shape = [self.y2 - self.y1, self.x2 - self.x1]
try:
tissueMask = glob.glob(stainfile + "/*_tissue_cut.tif")[0]
except IndexError:
try:
tissueMask = glob.glob(stainfile + "/*_tissue_cut_use.tif")[0]
except:
sys.stderr.write("mask file for tissue cut dose not exists, please check...")
sys.exit(1)
try:
pyramidImg = glob.glob(stainfile + "/*.rpi")[0]
except IOError:
sys.stderr.write("pyramidImg file does not exists, please check...")
exit(1)
try:
cellMaskFile = glob.glob(stainfile + "/*_mask.tif")[0]
if os.path.exists(cellMaskFile):
cellmask = tifffile.imread(cellMaskFile)
else:
cellmask = None
except IOError:
sys.stderr.write("cell mask file dose not exists")
except IndexError:
cellmask = None
sys.stderr.write("cell mask file dose not exists")
try:
registfile = glob.glob(stainfile + "/*_regist.tif")[0]
except IOError:
sys.stderr.write("regist image dose not exists, please check...")
# stainImg = cv2.imread(registfile, -1)
# stainImg = cv2.imread(registfile, -1)[cnt[1]: cnt[3], cnt[0]: cnt[2]]
# stainImg = cv2.flip(stainImg, 0)
# width, height = cnt[3] - cnt[1], cnt[2] - cnt[0]
# cmpImg = cv2.resize(stainImg, (width // 5, height // 5), cv2.INTER_CUBIC)
# cv2.imwrite(os.path.join(self.figpath, 'HE_{0}_{1}_{2}_{3}.png'.format(0, 0, height, width)), stainImg)
h5OutFile = os.path.join(self.figpath, self.snId + ".ssDNA.rpi")
try:
copyfile(pyramidImg, h5OutFile)
except IOError as e:
print("Unable to copy file. %s" % e)
exit(1)
except:
print("Unexpected error:", sys.exc_info())
exit(1)
# createPyramid(stainImg, h5OutFile, 0, 0)
mask = tifffile.imread(tissueMask)
# mask = cv2.imread(tissueMask, -1)
# mask = cv2.flip(mask, 0)
conArea = len(np.where(mask > 0)[0])
return conArea, mask, cellmask, 0, 0
def DnbMatting(self, binImg):
# kernel = np.array((
# [1, 1, 1, 1, 1],
# [1, 1, 1, 1, 1],
# [1, 1, 1, 1, 1],
# [1, 1, 1, 1, 1],
# [1, 1, 1, 1, 1]), dtype="float32")
# destImg = cv2.filter2D(binImg, -1, kernel)
gradx = cv2.Sobel(binImg, ddepth=-1, dx=1, dy=0, ksize=-1)
grady = cv2.Sobel(binImg, ddepth=-1, dx=0, dy=1, ksize=-1)
gradient = cv2.subtract(gradx, grady)
gradient = cv2.convertScaleAbs(gradient)
blurred = cv2.blur(gradient, (3, 3))
(_, thresh) = cv2.threshold(blurred, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (100, 100))
closed = cv2.morphologyEx(thresh, cv2.MORPH_CLOSE, kernel)
closed = cv2.erode(closed, None, iterations=10)
closed = cv2.dilate(closed, None, iterations=10)
(cnts, _) = cv2.findContours(closed.copy(), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
# imageFile = os.path.join(self.outpath, "{0}x{0}_image.tif".format(self.binSize))
if (len(cnts) < 1):
bx1 = self.x1
by1 = self.y1
bx2 = self.x2
by2 = self.y2
filterGene = self.genedf
# cv2.imwrite(imageFile, binImg)
else:
c = sorted(cnts, key=cv2.contourArea, reverse=True)[0]
rect = cv2.minAreaRect(c)
box = np.int0(cv2.boxPoints(rect))
# cv2.drawContours(binImg, [box], -1, (255, 0, 0), 1)
# cv2.imwrite(imageFile, binImg)
Xs = box[..., 0]
Ys = box[..., 1]
minX, maxX = max(0, min(Xs) - 10), max(Xs) + 10
minY, maxY = max(0, min(Ys) - 10), max(Ys) + 10
cv2.drawContours(binImg, [box], -1, (255, 0, 0), 1)
dnbMattingImgFile = os.path.join(self.outpath, "{0}X{0}_contour_image.png".format(self.binSize))
cv2.imwrite(dnbMattingImgFile, binImg)
bx1, bx2, by1, by2 = max(0, minX * self.binSize + self.x1), maxX * self.binSize + self.x1, max(0,
minY * self.binSize + self.y1), maxY * self.binSize + self.y1
filterGene = self.genedf.loc[
(self.genedf['x'] >= bx1) & (self.genedf['x'] <= bx2) & (self.genedf['y'] >= by1) & (
self.genedf['y'] <= by2)]
bx1 = filterGene['x'].min()
by1 = filterGene['y'].min()
filterGene['x'] = filterGene['x'] - bx1
filterGene['y'] = filterGene['y'] - by1
self.tissueOffsetX = bx1
self.tissueOffsetY = by1
self.genedf = filterGene
self.x1 = 0
self.y1 = 0
self.x2 = bx2 - bx1
self.y2 = by2 - by1
binImg = binImg[by1:by2, bx1:bx2]
self.ori_shape = (self.y2 - self.y1 + 1, self.x2 - self.x1 + 1)
# print ("minX: {0}\tmaxX: {1}\tminY: {2}\tmaxY: {3}".format(minX, maxX, minY, maxY))
print("x1: {0}\tx2: {1}\ty1: {2}\ty2: {3}".format(bx1, bx2, by1, by2))
filterGenefile = os.path.join(self.outpath, "{0}.gef".format(self.snId))
# with gzip.open(filterGenefile, 'wt') as writer:
# self.info="#FileFormat=GEMv0.1\n#SortedBy=None\n#BinSize=1\n#StereoChip={0}\n#OffsetX={1}\n#OffsetY={2}\n".format(self.snId, bx1, by1)
# writer.write(self.info)
# filterGene[['geneID', 'x', 'y', 'MIDCount']].to_csv(writer, index=None, sep="\t")
generate_bgef(self.infile, filterGenefile,stromics=self.omics,region=[bx1 - self.rawGefMinX, bx2 - self.rawGefMinX, by1 - self.rawGefMinY,by2 - self.rawGefMinY])
self.parseGefOffset(filterGenefile)
temppath = os.path.join(self.outpath, "dnb_merge")
os.makedirs(temppath, exist_ok=True)
gefplot.save_exp_heat_map(filterGenefile, os.path.join(temppath, "bin200.png"))
return bx1, by1, binImg
def FindContours(self, binImg, amp_shape, amp_factor, min_mean, platform, struct_kernel, low_thresh, high_thresh):
""" 抠图函数 """
mean_value = binImg.mean()
print("Image mean value: ", mean_value)
ampImg = self.AmpImg(binImg, amp_shape, amp_factor)
gradient = self.CalGradient(ampImg)
blurred = cv2.blur(gradient, (11, 11))
# thresholds = filters.threshold_isodata(blurred)
# _, threshold = cv2.threshold(blurred, thresholds, 255, cv2.THRESH_BINARY)
threshold = self.FindThreshold(blurred, low_thresh, high_thresh)
# cv2.imwrite(os.path.join(self.outpath, "Threshold_image.tif"), threshold)
if platform == 'SEQ500' or mean_value < 12:
struct_kernel = 25
closed = self.FillHoles(threshold, struct_kernel)
else:
# kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (5, 5))
# opening = cv2.erode(threshold, kernel)
closed = self.FillHoles(threshold, struct_kernel)
cnts, _ = cv2.findContours(closed, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
return cnts
# Dump tissue.gef
def dumpGef(self, outfile, df):
with h5py.File(outfile, 'w') as f:
# Sort by geneID, for generate /geneExp/bin1/gene
df['x'] += self.tissueOffsetX
df['y'] += self.tissueOffsetY
df = df.sort_values(by='geneID').reset_index(drop=True)
# minX = min(df['x'])
# minY = min(df['y'])
minX = self.snOffsetX
minY = self.snOffsetY
maxX = max(df['x'])
maxY = max(df['y'])
maxExp = max(df['MIDCount'])
df['x'] -= minX
df['y'] -= minY
data = df[['x', 'y', 'MIDCount']].to_records(index=False)
data.dtype.names = ('x', 'y', 'count')
middt = 'u1'
if maxExp < 255:
middt = 'u1'
elif maxExp < 65535:
middt = 'u2'
else:
middt = 'u4'
dt = np.dtype([('x', 'u4'), ('y', 'u4'), ('count', middt)])
dset = f.create_dataset("geneExp/bin1/expression", data=data, dtype=dt)
dset.attrs['minX'] = np.array([minX], dtype='u4')
dset.attrs['minY'] = np.array([minY], dtype='u4')
dset.attrs['maxX'] = np.array([maxX], dtype='u4')
dset.attrs['maxY'] = np.array([maxY], dtype='u4')
dset.attrs['maxExp'] = np.array([maxExp], dtype='u4')
dset.attrs['resolution'] = np.array(self.resolution, dtype='u4')
data = df[['geneID']].to_numpy()
offset = 0
genes = []
for k, g in groupby(data.flat):
gene = k
count = sum(1 for i in g)
genes.append((gene, offset, count))
offset += count
tid = h5py.h5t.C_S1.copy()
tid.set_size(32)
dt = np.dtype([('gene', tid), ('offset', 'u4'), ('count', 'u4')])
data = np.array(genes, dtype=dt)
dset = f.create_dataset("geneExp/bin1/gene", data=data)
# add version
f.attrs['version'] = np.array(self.version, dtype='u4')
# @profile
def Dumpresult(self, mask,maskLen):
""" merge结果 """
import time
t0 = time.time()
# glog.info(mask.shape)
# write bin1 to tissue.gef
dt = np.dtype([('x', 'u4'), ('y', 'u4'), ('count', 'u1')])
f = h5py.File(self.infile, 'r')
# f = h5py.File('/Users/berry/tmp/SS200000003BR_B3.tissue.gef', 'r')
self.rawGefMinX = f['/geneExp/bin1/expression'].attrs['minX'][0]
self.rawGefMinY = f['/geneExp/bin1/expression'].attrs['minY'][0]
self.rawGefMaxX = f['/geneExp/bin1/expression'].attrs['maxX'][0]
self.rawGefMaxY = f['/geneExp/bin1/expression'].attrs['maxY'][0]
exp = f['/geneExp/bin1/expression']
totalNum = exp.shape[0]
existExon = False
for k in f['/geneExp/bin1'].keys():
if ((f['/geneExp/bin1'][k].name) == '/geneExp/bin1/exon'):
existExon = True
break
if (existExon):
exon = f['/geneExp/bin1/exon']
exonDt=np.dtype('u1')
exonOutData = np.zeros(totalNum, dtype=exonDt)
# exonOutData = np.zeros(maskLen, dtype=exonDt)
# expOutData = np.zeros(maskLen, dtype=dt)
expOutData = np.zeros(totalNum, dtype=dt)
batch = totalNum // 4
iter = np.int64(0)
totalIter=np.int64(0)
gene = f['/geneExp/bin1/gene']
geneNp = np.array(gene.fields(['gene', 'offset', 'count'])[0:])
flag = np.zeros(geneNp.shape[0],dtype=np.int64)
tmpIter=0
for t in geneNp:
if tmpIter == 0:
flag[tmpIter] = t[2]
else:
flag[tmpIter] = t[2] + flag[tmpIter - 1]
tmpIter+=1
result=np.zeros(geneNp.shape[0],dtype=np.int64)
curIdx=0
for i in range(0, totalNum, batch):
endIdx = totalNum if i + batch > totalNum else i + batch
expNp = np.array(exp.fields(['x', 'y', 'count'])[i:endIdx])
if (existExon):
exonNp=np.array(exon[i:endIdx])
for ele in expNp:
self.total_umi+=ele[2]
if(mask[ele[1]][ele[0]]>0):
expOutData[iter]=ele
if (existExon):
exonOutData[iter]=exonNp[totalIter-i]
while(totalIter>=flag[curIdx]):
curIdx+=1
if curIdx>=geneNp.shape[0]:
break;
result[curIdx]+=1
iter += 1
totalIter+=1
glog.info("processed batch "+str(i//batch))
del expNp
if (existExon):
del exonNp
gc.collect()
outGenes=[]
offset=0
geneCount=0
for i in range(geneNp.shape[0]):
if(result[i]>0):
outGenes.append((geneNp[i][0],offset,result[i]))
offset+=result[i]
del geneNp
del result
del flag
if(self.omics=='Proteomics'):
tissueGef = os.path.join(self.outpath, "{0}.protein.tissue.gef".format(self.snId))
else:
tissueGef = os.path.join(self.outpath, "{0}.tissue.gef".format(self.snId))
h5out = h5py.File(tissueGef,'w')
dset=h5out.create_dataset('/geneExp/bin1/expression',data=expOutData[:iter],dtype=dt)
if(existExon):
if (self.omics!='Proteomics'):
exonSet=h5out.create_dataset('/geneExp/bin1/exon',data=exonOutData[:iter],dtype=exonDt)
maxExon=max(exonOutData[:iter])
exonSet.attrs['maxExon']=np.array([maxExon],dtype='u4')
minX = self.snOffsetX
minY = self.snOffsetY
maxX = max(expOutData['x'])
maxY = max(expOutData['y'])
maxExp = max(expOutData['count'])
dset.attrs['minX'] = np.array([minX], dtype='u4')
dset.attrs['minY'] = np.array([minY], dtype='u4')
dset.attrs['maxX'] = np.array([maxX], dtype='u4')
dset.attrs['maxY'] = np.array([maxY], dtype='u4')
dset.attrs['maxExp'] = np.array([maxExp], dtype='u4')
dset.attrs['resolution'] = np.array(self.resolution, dtype='u4')
del expOutData
if (existExon):
del exonOutData
gc.collect()
tid = h5py.h5t.TypeID.copy(h5py.h5t.C_S1)
tid.set_size(32)
geneDt = np.dtype([('gene', tid), ('offset', 'u4'), ('count', 'u4')])
geneOutNp = np.array(outGenes, dtype=geneDt)
genedset = h5out.create_dataset("geneExp/bin1/gene", data=geneOutNp, dtype=geneDt)
h5out.attrs['version'] = np.array(self.version, dtype='u4')
omicsTid = h5py.h5t.TypeID.copy(h5py.h5t.C_S1)
omicsTid.set_size(30)
h5out.attrs['omics'] = np.array(self.omics, dtype=omicsTid)
f.close()
h5out.close()
t1 = time.time()
print("Dumping result done. time used: {:.2f}".format(t1 - t0))
# return mergedf, coor, num_dnb
def DumpCellresult(self, maskImg, dnbdf):
""" 细胞分割结果 """
t0 = time.time()
"""
将二值mask，转为按细胞编号的32bit/64bit的label
"""
# _, labels = cv2.connectedComponents(maskImg)
num_labels, labels, stats, centroids = cv2.connectedComponentsWithStats(maskImg, connectivity=8)
cv2.imwrite(os.path.join(self.outpath, "cell_mask.tif"), labels)
"""
Merge 表达矩阵与细胞label
"""
tissuedf = pd.DataFrame()
dst = np.nonzero(labels)
tissuedf['x'] = dst[1]
tissuedf['y'] = dst[0]
tissuedf['label'] = labels[dst]
celldf = pd.merge(self.genedf, tissuedf, on=['x', 'y'], how='inner')
with gzip.open(os.path.join(self.outpath, "{0}.cell.gem.gz".format(self.snId)), 'wt') as writer:
writer.write(self.info)
celldf.to_csv(writer, sep='\t', index=False)
statDf = pd.DataFrame()
statDf['label'] = range(1, stats.shape[0])
statDf['x'] = stats[..., 1][1:]
statDf['y'] = stats[..., 0][1:]
statDf['CellArea'] = stats[..., 4][1:]
statDf.to_csv(os.path.join(self.outpath, "cell_stat.txt"), sep="\t", index=False)
t1 = time.time()
print("Dumping cell segmentation result done. time used: {:.2f}".format(t1 - t0))
"""
画cell bin统计图，并将cell bin的统计结果输出到log文件中
"""
celldf['Gene Number'] = celldf['geneID']
celldf['MID Count'] = celldf['MIDCount']
celldf['DNB Number'] = celldf['x'].astype(str) + "-" + celldf['y'].astype(str)
groupdf = celldf.groupby(['label'])
sum_info = groupdf[['MID Count']].sum()
count_info = groupdf[['Gene Number', 'DNB Number']].nunique()
statdf = pd.concat([sum_info, count_info], axis=1)
readsdf = pd.merge(celldf[['x', 'y', 'label']], dnbdf, how='inner')
groupreadsdf = readsdf['reads'].groupby(readsdf['label']).sum()
mReads = round(groupreadsdf.values.mean(), 2)
medReads = np.median(groupreadsdf.values)
logpath = os.path.join(self.outpath, "TissueCut.log")
log = open(logpath, "a")
self.gemStat(statdf, mReads, medReads, log, 0)
log.close()
def read_Geojson(self, geoFile):
"""
读取geojson，并转化为bit8的mask
"""
import geojson
mask = np.zeros(self.ori_shape, np.uint8)
with open(geoFile, "r") as geofile:
gj = geojson.load(geofile)
for i in gj['geometries']:
cv2.fillPoly(mask, np.array(i["coordinates"]), 255)
return mask
def TissueStat(self, mergedf, dnbdf, conArea, num_dnb, dnbreads, maxbinSize=200, step=50):
"""
结果统计
mergedf: 组织区域表达矩阵
coor：组织区域dnb坐标
dnbdf：组织区域 x, y, reads 对应表格
conArea：组织区域面积（dnb）
num_dnb: 组织下dnb数量
dnbreads: 总的reads数
maxbinSize/step: 输出某个binSize下的统计结果，default 200/50 即从1开始，输出步长为50的binning结果 （1，50，100，150，200）
统计结果包括基因数，UMI数，reads数等 以及对应的小提琴图/散点图
"""
num_gene = mergedf['geneID'].nunique()
num_umi = mergedf['MIDCount'].sum()
umiFraction = num_umi / self.total_umi * 100
print("Gene type: ", num_gene)
print("Total umi: ", num_umi)
# mergetis = pd.merge(coor, dnbdf, how='inner')
### fraction reads in spots under tissue
# dnbreads = dnbdf['reads'].sum()
print("DNB reads: ", dnbreads)
tisreads = dnbdf['reads'].sum()
print("Tissue area reads: ", tisreads)
if dnbreads != 0:
fraction = tisreads / dnbreads * 100
else:
fraction = 0
print("The fraction: ", fraction)
logpath = os.path.join(self.outpath, "TissueCut.log")
log = open(logpath, "w")
if self.stainUsed:
log.write("# Tissue Statistic Analysis with Stain Image\n")
else:
log.write("# Tissue Statistic Analysis\n")
log.write("Contour_Area\t{}\nNumber_of_DNB_Under_Tissue\t{}\nRatio\t{:.2f}% \n".format(conArea, num_dnb, (
num_dnb / conArea) * 100))
log.write("Total_Gene_Type\t{}\n".format(num_gene))
log.write("MID_Counts\t{}\n".format(num_umi))
log.write("Fraction_MID_in_Spots_Under_Tissue\t{:.2f}%\n".format(umiFraction))
log.write("Reads_Under_Tissue\t{}\n".format(tisreads))
log.write("Fraction_Reads_in_Spots_Under_Tissue\t{:.2f}%\n".format(fraction))
### Mean reads per spots/ median genes/ median UMI
for b in range(0, maxbinSize + 1, step):
if b == 0:
b = 1
df = self.binStat(mergedf, b)
_, meanReads, medReads = self._binStat(dnbdf, b, "reads")
self.gemStat(df, meanReads, medReads, log, b)
log.close()
def binStat(self, mergedf, binSize):
bindf = pd.DataFrame()
bindf['x'] = mergedf['x'] // binSize
bindf['y'] = mergedf['y'] // binSize
# bindf['Read Count'] = mergedf['reads']
bindf['Gene Number'] = mergedf['geneID']
bindf['MID Count'] = mergedf['MIDCount']
bindf['DNB Number'] = mergedf['x'].astype(str) + "-" + mergedf['y'].astype(str)
groupdf = bindf.groupby(['x', 'y'])
sum_info = groupdf[['MID Count']].sum()
count_info = groupdf[['Gene Number', 'DNB Number']].nunique()
statdf = pd.concat([sum_info, count_info], axis=1)
return statdf
def _binStat(self, mergedf, binSize, t):
bindf = pd.DataFrame()
bindf['x'] = mergedf['x'] // binSize
bindf['y'] = mergedf['y'] // binSize
bindf[t] = mergedf[t]
if t == 'geneID':
if binSize == 1:
groupdf = mergedf["geneID"].groupby([mergedf['x'], mergedf['y']]).count()
else:
groupdf = bindf[t].groupby([bindf['x'], bindf['y']]).nunique()
else:
groupdf = bindf[t].groupby([bindf['x'], bindf['y']]).sum()
if bindf.empty:
mean = 0
med = 0
else:
mean = round(groupdf.values.mean(), 2)
med = np.median(groupdf.values)
return groupdf, mean, med
def bin1GeneStat(self, mergedf):
groupdf = mergedf["geneID"].groupby([mergedf['x'], mergedf['y']]).count()
mean = round(groupdf.values.mean(), 3)
med = np.median(groupdf.values)
return groupdf, mean, med
def gemStat(self, df, mReads, medReads, log, b):
if (b != 1):
scapath = os.path.join(self.figpath, "scatter_{0}x{0}_MID_gene_counts.png".format(b if b != 0 else "cell"))
violinpath = os.path.join(self.figpath, "violin_{0}x{0}_MID_gene.png".format(b if b != 0 else "cell"))
statisticPath = os.path.join(self.figpath,
"statistic_{0}x{0}_MID_gene_DNB.png".format(b if b != 0 else "cell"))
plt.figure(figsize=(5, 5))
# sns.scatterplot(x=df['n_counts'], y=df['n_genes'], edgecolor="gray", color="gray")
plt.scatter(df['MID Count'], df['Gene Number'], color="gray", edgecolors="gray", s=0.8)
plt.grid()
plt.xlabel("MID Count")