-
Notifications
You must be signed in to change notification settings - Fork 31
/
Copy pathlayer_specificity.R
executable file
·1828 lines (1642 loc) · 60.5 KB
/
layer_specificity.R
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
library('SingleCellExperiment')
library('here')
library('jaffelab')
library('scater')
library('scran')
library('pheatmap')
library('readxl')
library('Polychrome')
library('cluster')
library('limma')
library('sessioninfo')
dir.create('pdf', showWarnings = FALSE)
dir.create('rda', showWarnings = FALSE)
## Load data
load(here(
'Analysis',
'Human_DLPFC_Visium_processedData_sce_scran.Rdata'
))
## For plotting
source(here('Analysis', 'spatialLIBD_global_plot_code.R'))
genes <- paste0(rowData(sce)$gene_name, '; ', rowData(sce)$gene_id)
## Functions derived from this script, to make it easier to resume the work
sce_layer_file <-
here('Analysis', 'Layer_Guesses', 'rda', 'sce_layer.Rdata')
if (file.exists(sce_layer_file))
load(sce_layer_file, verbose = TRUE)
source(here('Analysis', 'Layer_Guesses', 'layer_specificity_functions.R'))
## Load layer guesses
load(here('Analysis', 'Layer_Guesses', 'rda',
'layer_guess_tab.Rdata'))
## Add layer guesses to the sce object
sce$layer_guess <- NA
m <- match(sce$key, layer_guess_tab$key)
table(is.na(m))
# FALSE TRUE
# 47329 352
sce$layer_guess[!is.na(m)] <- layer_guess_tab$layer[m[!is.na(m)]]
## Check layer guesses across images
options(width = 120)
with(colData(sce), addmargins(table(layer_guess, sample_name, useNA = 'ifany')))
# sample_name
# layer_guess 151507 151508 151509 151510 151669 151670 151671 151672 151673 151674 151675 151676 Sum
# Layer 1 817 866 1189 1180 0 0 0 0 273 380 328 289 5322
# Layer 2 305 295 602 650 0 0 0 0 253 224 275 254 2858
# Layer 2/3 0 0 0 0 2141 2175 1918 1575 0 0 0 0 7809
# Layer 3 1215 1385 1884 1774 0 0 0 0 989 924 771 836 9778
# Layer 4 369 373 369 318 364 211 245 304 218 247 275 254 3547
# Layer 5 675 737 363 310 510 581 721 728 673 621 732 649 7300
# Layer 6 486 525 215 179 391 308 760 882 692 614 533 616 6201
# WM 354 200 166 184 230 209 449 399 513 625 652 533 4514
# <NA> 5 3 1 39 25 14 17 127 28 38 26 29 352
# Sum 4226 4384 4789 4634 3661 3498 4110 4015 3639 3673 3592 3460 47681
## Drop the layer guess NAs for now
sce_original <- sce
sce <- sce[, !is.na(sce$layer_guess)]
dim(sce)
# [1] 33538 47329
## Next, re-label "Layer 2/3" as "Layer 3" for now
## (there's more layer 3 in the other samples than 2 anyway)
sce$layer_guess[sce$layer_guess == 'Layer 2/3'] <- 'Layer 3'
## Make it into a factor with WM as the reference
## and remove spaces
sce$layer_guess <-
factor(gsub(' ', '', sce$layer_guess), levels = c('WM', paste0('Layer', 1:6)))
## Check again
with(colData(sce), addmargins(table(layer_guess, sample_name, useNA = 'ifany')))
# sample_name
# layer_guess 151507 151508 151509 151510 151669 151670 151671 151672 151673 151674 151675 151676 Sum
# WM 354 200 166 184 230 209 449 399 513 625 652 533 4514
# Layer1 817 866 1189 1180 0 0 0 0 273 380 328 289 5322
# Layer2 305 295 602 650 0 0 0 0 253 224 275 254 2858
# Layer3 1215 1385 1884 1774 2141 2175 1918 1575 989 924 771 836 17587
# Layer4 369 373 369 318 364 211 245 304 218 247 275 254 3547
# Layer5 675 737 363 310 510 581 721 728 673 621 732 649 7300
# Layer6 486 525 215 179 391 308 760 882 692 614 533 616 6201
# Sum 4221 4381 4788 4595 3636 3484 4093 3888 3611 3635 3566 3431 47329
## Split spots by layer and image
## Based on here('Analysis', 'sce_scran.R')
## /dcs04/lieber/lcolladotor/with10x_LIBD001/HumanPilot/Analysis/sce_scran.R
layerIndexes <-
splitit(paste0(sce$sample_name, '_', sce$layer_guess))
## Collapse UMIs
umiComb <-
sapply(layerIndexes, function(ii)
rowSums(assays(sce)$counts[, ii, drop = FALSE]))
dim(umiComb)
# [1] 33538 76
## That's because 2 layers are absent in subject 2
stopifnot(12 * 7 - 2 * 4 == ncol(umiComb))
## Build a data.frame with the pheno data by layer
layer_df <- data.frame(
sample_name = ss(colnames(umiComb), '_', 1),
layer_guess = factor(ss(colnames(umiComb), '_', 2), levels = levels(sce$layer_guess)),
stringsAsFactors = FALSE
)
m_layer <- match(layer_df$sample_name, sce$sample_name)
layer_df$subject <- sce$subject[m_layer]
layer_df$position <- sce$position[m_layer]
layer_df$replicate <- sce$replicate[m_layer]
layer_df$subject_position <- sce$subject_position[m_layer]
rownames(layer_df) <- colnames(umiComb)
## Get a sample-specific size factors, instead of sample/layer size factors
umiComb_sample <-
sapply(splitit(ss(colnames(umiComb), '_', 1)), function(ii) {
rowSums(umiComb[, ii, drop = FALSE])
})
umiComb_sample_size_fac <- librarySizeFactors(umiComb_sample)
umiComb_sample_size_fac_layer <-
rep(umiComb_sample_size_fac, lengths(splitit(ss(
colnames(umiComb), '_', 1
))))
names(umiComb_sample_size_fac_layer) <- colnames(umiComb)
## Compare the size factors
pdf('pdf/size_factors_comparison.pdf', useDingbats = FALSE)
plot(
librarySizeFactors(umiComb),
umiComb_sample_size_fac_layer,
bg = Polychrome::palette36.colors(7)[as.integer(layer_df$layer_guess)],
pch = 21,
xlab = 'Default size factors',
ylab = 'Sample size factors (repeated across layers)',
main = 'Compare size factors'
)
legend(
x = 1,
y = 1.5,
legend = levels(layer_df$layer_guess),
col = Polychrome::palette36.colors(7),
lwd = 3,
bty = 'n',
ncol = 2
)
dev.off()
## There are layer differences, so it's likely best to use the default
## size factors instead
## Build a new sce object
sce_layer <-
logNormCounts(SingleCellExperiment(
list(counts = umiComb),
colData = layer_df,
rowData = rowData(sce)
))
# ,size_factors = umiComb_sample_size_fac_layer)
## From Lukas's code
## https://github.com/LieberInstitute/HumanPilot/commit/0bd87cb5cd69863ceafc343f5456dcafca0dccb6#diff-6c1a4dea06c53f935341cef1d3bccbfdR158
## for identifying the mitochondrial genes
ix_mito <- grep("^MT-", rowData(sce_layer)$gene_name)
## Several of the mitochondrial genes are among the most expressed ones
pdf('pdf/highest_expressed.pdf', useDingbats = FALSE)
plotHighestExprs(
sce_layer,
exprs_values = "counts",
colour_cells_by = 'layer_guess',
feature_names_to_plot = 'gene_name',
) + ggplot2::ggtitle('All genes') + ggplot2::xlab('Percent among counts') +
ggplot2::scale_colour_manual(values = unname(Polychrome::palette36.colors(7)),
name = 'Layer')
plotHighestExprs(
sce_layer,
exprs_values = "counts",
colour_cells_by = 'layer_guess',
feature_names_to_plot = 'gene_name',
drop_features = seq_len(nrow(sce_layer))[-ix_mito]
) + ggplot2::ggtitle('Only mitochondrial genes') +
ggplot2::xlab('Percent among counts') +
ggplot2::scale_colour_manual(values = unname(Polychrome::palette36.colors(7)),
name = 'Layer')
plotHighestExprs(
sce_layer,
exprs_values = "counts",
colour_cells_by = 'layer_guess',
feature_names_to_plot = 'gene_name',
drop_features = ix_mito,
) + ggplot2::ggtitle('Without mitochondrial genes') +
ggplot2::xlab('Percent among counts') +
ggplot2::scale_colour_manual(values = unname(Polychrome::palette36.colors(7)),
name = 'Layer')
dev.off()
pdf('pdf/mt_expression.pdf', useDingbats = FALSE)
## Hack my way around it so I can show the gene symbols
x <-
plotExpression(sce_layer, rownames(sce_layer)[ix_mito], x = 'layer_guess')
x$data$Feature <-
rowData(sce_layer)$gene_name[match(as.character(x$data$Feature), rownames(sce_layer))]
print(x)
## Different yet similar hack
x <-
plotExpression(sce_layer, rownames(sce_layer)[ix_mito], colour_by = 'layer_guess')
x$data$X <-
rowData(sce_layer)$gene_name[match(as.character(x$data$X), rownames(sce_layer))]
print(x +
ggplot2::scale_fill_manual(
values = unname(Polychrome::palette36.colors(7)),
name = 'Layer'
))
dev.off()
rm(x)
## Drop mitochondrial genes
sce_layer <- sce_layer[-ix_mito, ]
## Find which genes to drop due to low expression values
sce_layer_avg <- calculateAverage(sce_layer)
summary(sce_layer_avg)
# Min. 1st Qu. Median Mean 3rd Qu. Max.
# 0.000 0.007 1.186 53.215 31.427 9507.091
## Results without dropping the mitochondrial genes:
# Min. 1st Qu. Median Mean 3rd Qu. Max.
# 0.00 0.01 1.19 64.31 31.50 59084.15
sce_layer_avg_logcounts <-
calculateAverage(sce_layer, exprs_values = 'logcounts')
summary(sce_layer_avg_logcounts)
# Min. 1st Qu. Median Mean 3rd Qu. Max.
# 0.000000 0.003973 1.379983 4.770529 9.678251 26.245639
## Results without dropping the mitochondrial genes:
# Min. 1st Qu. Median Mean 3rd Qu. Max.
# 0.000000 0.003973 1.384647 4.779067 9.683280 30.899169
# hist(sce_layer_avg)
# hist(sce_layer_avg_logcounts, breaks = 50)
addmargins(
table(
'avg log > 0.1' = sce_layer_avg_logcounts > 0.1,
'avg log > 0.05' = sce_layer_avg_logcounts > 0.05,
'avg log > 0' = sce_layer_avg_logcounts > 0
)
)
# , , avg log > 0 = FALSE
#
# avg log > 0.05
# avg log > 0.1 FALSE TRUE Sum
# FALSE 7943 0 7943
# TRUE 0 0 0
# Sum 7943 0 7943
#
# , , avg log > 0 = TRUE
#
# avg log > 0.05
# avg log > 0.1 FALSE TRUE Sum
# FALSE 2746 1015 3761
# TRUE 0 21821 21821
# Sum 2746 22836 25582
#
# , , avg log > 0 = Sum
#
# avg log > 0.05
# avg log > 0.1 FALSE TRUE Sum
# FALSE 10689 1015 11704
# TRUE 0 21821 21821
# Sum 10689 22836 33525
## From scater's vignette at
## http://bioconductor.org/packages/release/bioc/vignettes/scater/inst/doc/overview.html#333_subsetting_by_row
keep_feature <- nexprs(sce_layer, byrow = TRUE) > 0
## The avg log counts cutoff would be more strict
addmargins(table(keep_feature, 'avg log > 0.05' = sce_layer_avg_logcounts > 0.05))
# avg log > 0.05
# keep_feature FALSE TRUE Sum
# FALSE 7943 0 7943
# TRUE 2746 22836 25582
# Sum 10689 22836 33525
## From scater's vignette:
# detected: the percentage of cells with non-zero counts for each gene.
per.feat <- perFeatureQCMetrics(sce_layer)
summary(per.feat$detected)
# Min. 1st Qu. Median Mean 3rd Qu. Max.
# 0.000 1.316 55.263 51.674 100.000 100.000
addmargins(
table(
'Detected >5%' = per.feat$detected > 5,
'avg log > 0.05' = sce_layer_avg_logcounts > 0.05,
keep_feature
)
)
# , , keep_feature = FALSE
#
# avg log > 0.05
# Detected >5% FALSE TRUE Sum
# FALSE 7943 0 7943
# TRUE 0 0 0
# Sum 7943 0 7943
#
# , , keep_feature = TRUE
#
# avg log > 0.05
# Detected >5% FALSE TRUE Sum
# FALSE 2389 505 2894
# TRUE 357 22331 22688
# Sum 2746 22836 25582
#
# , , keep_feature = Sum
#
# avg log > 0.05
# Detected >5% FALSE TRUE Sum
# FALSE 10332 505 10837
# TRUE 357 22331 22688
# Sum 10689 22836 33525
pdf('pdf/avg_logcounts_VS_detected_percent.pdf', useDingbats = FALSE)
plot(
sce_layer_avg_logcounts,
log10(per.feat$detected),
bg = add.alpha('black', 1 / 10),
pch = 21,
col = NA,
xlab = 'Average logcounts',
ylab = 'Layer/image % detected (log10)'
)
abline(v = 0.05,
col = 'red',
lwd = 2,
lty = 2)
abline(
h = log10(5),
col = 'blue',
lwd = 2,
lty = 2
)
plot(
sce_layer_avg_logcounts,
per.feat$detected,
xlim = c(0, 0.1),
ylim = c(0, 10),
bg = add.alpha('black', 1 / 10),
pch = 21,
col = NA,
xlab = 'Average logcounts',
ylab = 'Layer/image % detected'
)
abline(v = 0.05,
col = 'red',
lwd = 2,
lty = 2)
abline(h = 5,
col = 'blue',
lwd = 2,
lty = 2)
plot(
sce_layer_avg_logcounts,
log10(per.feat$detected),
xlim = c(0.05, max(sce_layer_avg_logcounts)),
ylim = log10(c(5, 100)),
bg = add.alpha('black', 1 / 10),
pch = 21,
col = NA,
xlab = 'Average logcounts',
ylab = 'Layer/image % detected (log10)'
)
abline(v = 0.05,
col = 'red',
lwd = 2,
lty = 2)
abline(
h = log10(5),
col = 'blue',
lwd = 2,
lty = 2
)
dev.off()
## The 5% detected means that the gene must be expressed in at least 4/76 layer/image combinations
4 / 76
# [1] 0.05263158
## Drop genes
selected_genes <-
which(per.feat$detected > 5 & sce_layer_avg_logcounts > 0.05)
length(selected_genes)
# [1] 22331
sce_layer <- sce_layer[selected_genes,]
dim(sce_layer)
# [1] 22331 76
## For comparison, try with the other library size factors
sce_layer_sfac <-
logNormCounts(SingleCellExperiment(
list(counts = umiComb),
colData = layer_df,
rowData = rowData(sce)
),
size_factors = umiComb_sample_size_fac_layer)[-ix_mito, ][selected_genes,]
## From scater's vignette at
## http://bioconductor.org/packages/release/bioc/vignettes/scater/inst/doc/overview.html#34_variable-level_qc
vars <- getVarianceExplained(
sce_layer,
variables = c(
'sample_name',
'layer_guess',
'subject',
'subject_position',
'position'
)
)
vars_sfac <- getVarianceExplained(
sce_layer_sfac,
variables = c(
'sample_name',
'layer_guess',
'subject',
'subject_position',
'position'
)
)
pdf('pdf/gene_explanatory_vars.pdf', useDingbats = FALSE)
plotExplanatoryVariables(vars) + ggplot2::ggtitle('With default sample size factors; no MT')
## Hack it so the colors are the same in both plots
plotExplanatoryVariables(vars_sfac) +
ggplot2::ggtitle('With sample size factors (repeated across layers); no MT') +
ggplot2::scale_colour_manual(values = unname(scater:::.get_palette('tableau10medium')[c(2, 1, 3:5)]),
name = '')
dev.off()
## From the above plots it indeed looks like using the default size factors is best
## From here('Analysis', 'sce_scran.R')
## Find the highly variable genes
dec <- modelGeneVar(sce_layer,
block = sce_layer$subject_position)
pdf('pdf/modelGeneVar.pdf', useDingbats = FALSE)
mapply(function(block, blockname) {
plot(
block$mean,
block$total,
xlab = "Mean log-expression",
ylab = "Variance",
main = blockname
)
curve(metadata(block)$trend(x),
col = "blue",
add = TRUE)
}, dec$per.block, names(dec$per.block))
dev.off()
top.hvgs <- getTopHVGs(dec, prop = 0.1)
length(top.hvgs)
# [1] 1280
## Run PCA with a reduced number of components to avoid errors
## (since we are starting with 76 columns anyway instead of thousands)
sce_layer <-
runPCA(sce_layer, subset_row = top.hvgs, ncomponents = 20)
## Default perplexity is 15
mat <-
scater:::.get_mat_from_sce(
sce_layer,
exprs_values = 'logcounts',
dimred = 'PCA',
n_dimred = NULL
)
dim(mat)
min(50, floor(nrow(mat) / 5))
# [1] 15
## Run TSNE and UMAP
set.seed(20200121)
sce_layer <-
runTSNE(sce_layer,
dimred = 'PCA',
name = 'TSNE_perplexity5',
perplexity = 5)
set.seed(20200121)
sce_layer <-
runTSNE(sce_layer,
dimred = 'PCA',
name = 'TSNE_perplexity15',
perplexity = 15)
set.seed(20200121)
sce_layer <-
runTSNE(sce_layer,
dimred = 'PCA',
name = 'TSNE_perplexity20',
perplexity = 20)
set.seed(20200121)
sce_layer <-
runUMAP(sce_layer, dimred = 'PCA', name = 'UMAP_neighbors15')
sce_layer$layer_guess_reordered <-
factor(sce_layer$layer_guess, levels = c(paste0('Layer', 1:6), 'WM'))
pdf('pdf/reduced_dim_PCA.pdf', useDingbats = FALSE)
plotReducedDim(
sce_layer,
dimred = 'PCA',
colour_by = 'layer_guess_reordered',
theme_size = 20,
point_size = 5
) +
ggplot2::scale_fill_manual(values = unname(Polychrome::palette36.colors(7))[c(2:7, 1)],
name = 'Layer')
plotReducedDim(
sce_layer,
dimred = 'PCA',
colour_by = 'sample_name',
theme_size = 20,
point_size = 5
)
plotReducedDim(
sce_layer,
dimred = 'PCA',
colour_by = 'subject_position',
theme_size = 20,
point_size = 5
)
plotReducedDim(
sce_layer,
dimred = 'PCA',
colour_by = 'subject',
theme_size = 20,
point_size = 5
)
plotReducedDim(
sce_layer,
dimred = 'PCA',
colour_by = 'position',
theme_size = 20,
point_size = 5
)
dev.off()
## Just to compare the PCA on the data using library size factors
## by sample repeated across layers
sce_layer_sfac <-
runPCA(sce_layer_sfac,
subset_row = top.hvgs,
ncomponents = 20)
pdf('pdf/reduced_dim_PCA_sfact.pdf', useDingbats = FALSE)
plotReducedDim(
sce_layer_sfac,
dimred = 'PCA',
colour_by = 'layer_guess',
theme_size = 20
) +
ggplot2::scale_fill_manual(values = unname(Polychrome::palette36.colors(7)),
name = 'Layer')
plotReducedDim(
sce_layer_sfac,
dimred = 'PCA',
colour_by = 'sample_name',
theme_size = 20
)
plotReducedDim(
sce_layer_sfac,
dimred = 'PCA',
colour_by = 'subject_position',
theme_size = 20
)
plotReducedDim(
sce_layer_sfac,
dimred = 'PCA',
colour_by = 'subject',
theme_size = 20
)
plotReducedDim(
sce_layer_sfac,
dimred = 'PCA',
colour_by = 'position',
theme_size = 20
)
dev.off()
## Comparing the two, it seems to me that the default library size reduces the
## layer differences but makes more sense in the PCA plot:
## like it's PC1 for WM vs non-WM, then PC2 across layers 1 through 6
## instead of PC1 for some layers vs others, PC2 for WM vs layer 2?
## with layers like 6 and 5 across the PC1 vs PC2 plot.
pdf('pdf/reduced_dim_TSNE_perplexity5.pdf', useDingbats = FALSE)
plotReducedDim(
sce_layer,
dimred = 'TSNE_perplexity5',
colour_by = 'layer_guess_reordered',
theme_size = 20,
point_size = 5
) +
ggplot2::scale_fill_manual(values = unname(Polychrome::palette36.colors(7))[c(2:7, 1)],
name = 'Layer')
plotReducedDim(
sce_layer,
dimred = 'TSNE_perplexity5',
colour_by = 'sample_name',
theme_size = 20,
point_size = 5
)
plotReducedDim(
sce_layer,
dimred = 'TSNE_perplexity5',
colour_by = 'subject_position',
theme_size = 20,
point_size = 5
)
plotReducedDim(
sce_layer,
dimred = 'TSNE_perplexity5',
colour_by = 'subject',
theme_size = 20,
point_size = 5
)
plotReducedDim(
sce_layer,
dimred = 'TSNE_perplexity5',
colour_by = 'position',
theme_size = 20,
point_size = 5
)
dev.off()
pdf('pdf/reduced_dim_TSNE_perplexity15.pdf', useDingbats = FALSE)
plotReducedDim(
sce_layer,
dimred = 'TSNE_perplexity15',
colour_by = 'layer_guess_reordered',
theme_size = 20,
point_size = 5
) +
ggplot2::scale_fill_manual(values = unname(Polychrome::palette36.colors(7))[c(2:7, 1)],
name = 'Layer')
plotReducedDim(
sce_layer,
dimred = 'TSNE_perplexity15',
colour_by = 'sample_name',
theme_size = 20,
point_size = 5
)
plotReducedDim(
sce_layer,
dimred = 'TSNE_perplexity15',
colour_by = 'subject_position',
theme_size = 20,
point_size = 5
)
plotReducedDim(
sce_layer,
dimred = 'TSNE_perplexity15',
colour_by = 'subject',
theme_size = 20,
point_size = 5
)
plotReducedDim(
sce_layer,
dimred = 'TSNE_perplexity15',
colour_by = 'position',
theme_size = 20,
point_size = 5
)
dev.off()
pdf('pdf/reduced_dim_TSNE_perplexity20.pdf', useDingbats = FALSE)
plotReducedDim(
sce_layer,
dimred = 'TSNE_perplexity20',
colour_by = 'layer_guess_reordered',
theme_size = 20,
point_size = 5
) +
ggplot2::scale_fill_manual(values = unname(Polychrome::palette36.colors(7))[c(2:7, 1)],
name = 'Layer')
plotReducedDim(
sce_layer,
dimred = 'TSNE_perplexity20',
colour_by = 'sample_name',
theme_size = 20,
point_size = 5
)
plotReducedDim(
sce_layer,
dimred = 'TSNE_perplexity20',
colour_by = 'subject_position',
theme_size = 20,
point_size = 5
)
plotReducedDim(
sce_layer,
dimred = 'TSNE_perplexity20',
colour_by = 'subject',
theme_size = 20,
point_size = 5
)
plotReducedDim(
sce_layer,
dimred = 'TSNE_perplexity20',
colour_by = 'position',
theme_size = 20,
point_size = 5
)
dev.off()
pdf('pdf/reduced_dim_UMAP_neighbors15.pdf', useDingbats = FALSE)
plotReducedDim(
sce_layer,
dimred = 'UMAP_neighbors15',
colour_by = 'layer_guess_reordered',
theme_size = 20,
point_size = 5
) +
ggplot2::scale_fill_manual(values = unname(Polychrome::palette36.colors(7))[c(2:7, 1)],
name = 'Layer')
plotReducedDim(
sce_layer,
dimred = 'UMAP_neighbors15',
colour_by = 'sample_name',
theme_size = 20,
point_size = 5
)
plotReducedDim(
sce_layer,
dimred = 'UMAP_neighbors15',
colour_by = 'subject_position',
theme_size = 20,
point_size = 5
)
plotReducedDim(
sce_layer,
dimred = 'UMAP_neighbors15',
colour_by = 'subject',
theme_size = 20,
point_size = 5
)
plotReducedDim(
sce_layer,
dimred = 'UMAP_neighbors15',
colour_by = 'position',
theme_size = 20,
point_size = 5
)
dev.off()
## Across PCA, TSNE and UMAP the WM for samples 151669 and 151670
## (middle subject, position 0) group closer to the layer 6 data from
## the other images.
## In TSNE and UMAP, there points do group by subject more than they
## do in PCA space.
set.seed(20200121)
sced <-
denoisePCA(
sce_layer,
dec,
subset.row = top.hvgs,
max.rank = 20,
min.rank = 2
)
ncol(reducedDim(sced, "PCA"))
# [1] 7
## Unlike here('Analysis', 'sce_scran.R') here this code ran within 2-5 secs
choices <-
getClusteredPCs(reducedDim(sce_layer),
max.rank = 20,
min.rank = 2)
npcs <- metadata(choices)$chosen
npcs
# [1] 5
reducedDim(sce_layer, "PCAsub") <-
reducedDim(sce_layer, "PCA")[, seq_len(npcs), drop = FALSE]
## This plot doesn't seem very informative in this scenario...
pdf('pdf/PC_choices.pdf', useDingbats = FALSE)
plot(choices$n.pcs,
choices$n.clusters,
xlab = "Number of PCs",
ylab = "Number of clusters")
abline(a = 1, b = 1, col = "red")
abline(v = metadata(choices)$chosen,
col = "grey80",
lty = 2)
dev.off()
g_k5 <- buildSNNGraph(sce_layer, k = 5, use.dimred = 'PCA')
g_walk_k5 <- igraph::cluster_walktrap(g_k5)
clust_k5 <- sort_clusters(g_walk_k5$membership)
save(g_k5, g_walk_k5, file = 'rda/g_k5.Rdata')
addmargins(table(clust_k5, sce_layer$layer_guess))
# clust_k5 WM Layer 1 Layer 2 Layer 3 Layer 4 Layer 5 Layer 6 Sum
# 1 0 8 8 12 0 0 0 28
# 2 0 0 0 0 12 12 0 24
# 3 3 0 0 0 0 0 12 15
# 4 9 0 0 0 0 0 0 9
# Sum 12 8 8 12 12 12 12 76
## ehem... that's why it makes more sense to have k = 7 :P
## since we have 7 layers. That was the k that denoisePCA()
## suggested earlier
g_k7 <- buildSNNGraph(sce_layer, k = 7, use.dimred = 'PCA')
g_walk_k7 <- igraph::cluster_walktrap(g_k7)
clust_k7 <- sort_clusters(g_walk_k7$membership)
save(g_k7, g_walk_k7, file = 'rda/g_k7.Rdata')
addmargins(table(clust_k7, sce_layer$layer_guess))
# clust_k7 WM Layer 1 Layer 2 Layer 3 Layer 4 Layer 5 Layer 6 Sum
# 1 0 8 8 12 10 0 0 38
# 2 4 0 0 0 2 12 12 30
# 3 8 0 0 0 0 0 0 8
# Sum 12 8 8 12 12 12 12 76
## Hm... cut at k = 7
clust_k5_k7 <- sort_clusters(igraph::cut_at(g_walk_k5, n = 7))
clust_k7_k7 <- sort_clusters(igraph::cut_at(g_walk_k7, n = 7))
addmargins(table(clust_k5_k7, sce_layer$layer_guess))
# clust_k5_k7 WM Layer 1 Layer 2 Layer 3 Layer 4 Layer 5 Layer 6 Sum
# 1 0 0 8 12 0 0 0 20
# 2 0 0 0 0 4 12 0 16
# 3 3 0 0 0 0 0 12 15
# 4 9 0 0 0 0 0 0 9
# 5 0 8 0 0 0 0 0 8
# 6 0 0 0 0 4 0 0 4
# 7 0 0 0 0 4 0 0 4
# Sum 12 8 8 12 12 12 12 76
addmargins(table(clust_k7_k7, sce_layer$layer_guess))
# clust_k7_k7 WM Layer 1 Layer 2 Layer 3 Layer 4 Layer 5 Layer 6 Sum
# 1 0 0 4 12 0 0 0 16
# 2 0 0 0 0 2 12 0 14
# 3 0 1 0 0 10 0 0 11
# 4 0 7 4 0 0 0 0 11
# 5 2 0 0 0 0 0 8 10
# 6 8 0 0 0 0 0 0 8
# 7 2 0 0 0 0 0 4 6
# Sum 12 8 8 12 12 12 12 76
## Try with all 20 PCs
g_k20 <- buildSNNGraph(sce_layer, k = 20, use.dimred = 'PCA')
g_walk_k20 <- igraph::cluster_walktrap(g_k20)
clust_k20 <- sort_clusters(g_walk_k20$membership)
save(g_k20, g_walk_k20, file = 'rda/g_k20.Rdata')
clust_k20_k7 <- sort_clusters(igraph::cut_at(g_walk_k20, n = 7))
addmargins(table(clust_k20, sce_layer$layer_guess))
# clust_k20 WM Layer 1 Layer 2 Layer 3 Layer 4 Layer 5 Layer 6 Sum
# 1 0 8 8 12 12 0 0 40
# 2 12 0 0 0 0 12 12 36
# Sum 12 8 8 12 12 12 12 76
addmargins(table(clust_k20_k7, sce_layer$layer_guess))
# clust_k20_k7 WM Layer 1 Layer 2 Layer 3 Layer 4 Layer 5 Layer 6 Sum
# 1 6 0 0 0 0 0 11 17
# 2 0 5 1 11 0 0 0 17
# 3 1 0 0 0 0 12 1 14
# 4 0 1 0 0 12 0 0 13
# 5 0 0 7 0 0 0 0 7
# 6 5 0 0 0 0 0 0 5
# 7 0 2 0 1 0 0 0 3
# Sum 12 8 8 12 12 12 12 76
## They are all highly variable
addmargins(table(clust_k7_k7, clust_k5_k7))
# clust_k5_k7
# clust_k7_k7 1 2 3 4 5 6 7 Sum
# 1 16 0 0 0 0 0 0 16
# 2 0 14 0 0 0 0 0 14
# 3 0 2 0 0 1 4 4 11
# 4 4 0 0 0 7 0 0 11
# 5 0 0 10 0 0 0 0 10
# 6 0 0 0 8 0 0 0 8
# 7 0 0 5 1 0 0 0 6
# Sum 20 16 15 9 8 4 4 76
addmargins(table(clust_k7_k7, clust_k20_k7))
# clust_k20_k7
# clust_k7_k7 1 2 3 4 5 6 7 Sum
# 1 0 12 0 0 3 0 1 16
# 2 0 0 12 2 0 0 0 14
# 3 0 0 0 11 0 0 0 11
# 4 0 5 0 0 4 0 2 11
# 5 10 0 0 0 0 0 0 10
# 6 3 0 0 0 0 5 0 8
# 7 4 0 2 0 0 0 0 6
# Sum 17 17 14 13 7 5 3 76
## Hm... could it be that it's picking up the subject variability?
addmargins(table(clust_k7_k7, sce_layer$subject))
# clust_k7_k7 Br5292 Br5595 Br8100 Sum
# 1 4 4 8 16
# 2 4 6 4 14
# 3 4 2 5 11
# 4 8 0 3 11
# 5 4 6 0 10
# 6 4 0 4 8
# 7 0 2 4 6
# Sum 28 20 28 76
addmargins(table(clust_k7_k7, sce_layer$layer_guess, sce_layer$subject))
# , , = Br5292
#
#
# clust_k7_k7 WM Layer 1 Layer 2 Layer 3 Layer 4 Layer 5 Layer 6 Sum
# 1 0 0 0 4 0 0 0 4
# 2 0 0 0 0 0 4 0 4
# 3 0 0 0 0 4 0 0 4
# 4 0 4 4 0 0 0 0 8
# 5 0 0 0 0 0 0 4 4
# 6 4 0 0 0 0 0 0 4
# 7 0 0 0 0 0 0 0 0
# Sum 4 4 4 4 4 4 4 28
#
# , , = Br5595
#
#
# clust_k7_k7 WM Layer 1 Layer 2 Layer 3 Layer 4 Layer 5 Layer 6 Sum
# 1 0 0 0 4 0 0 0 4
# 2 0 0 0 0 2 4 0 6
# 3 0 0 0 0 2 0 0 2
# 4 0 0 0 0 0 0 0 0
# 5 2 0 0 0 0 0 4 6
# 6 0 0 0 0 0 0 0 0
# 7 2 0 0 0 0 0 0 2
# Sum 4 0 0 4 4 4 4 20
#
# , , = Br8100
#
#
# clust_k7_k7 WM Layer 1 Layer 2 Layer 3 Layer 4 Layer 5 Layer 6 Sum
# 1 0 0 4 4 0 0 0 8
# 2 0 0 0 0 0 4 0 4
# 3 0 1 0 0 4 0 0 5
# 4 0 3 0 0 0 0 0 3
# 5 0 0 0 0 0 0 0 0
# 6 4 0 0 0 0 0 0 4
# 7 0 0 0 0 0 0 4 4
# Sum 4 4 4 4 4 4 4 28
#
# , , = Sum
#
#
# clust_k7_k7 WM Layer 1 Layer 2 Layer 3 Layer 4 Layer 5 Layer 6 Sum
# 1 0 0 4 12 0 0 0 16
# 2 0 0 0 0 2 12 0 14
# 3 0 1 0 0 10 0 0 11
# 4 0 7 4 0 0 0 0 11
# 5 2 0 0 0 0 0 8 10
# 6 8 0 0 0 0 0 0 8
# 7 2 0 0 0 0 0 4 6
# Sum 12 8 8 12 12 12 12 76
## So subject 1, it merges layers 1 & 2
## subject 2, it breaks layer 4 (mixes it a bit with layer 5) and WM
## subject 3, for one it merges layer 1 with 4, merges layers 2 and 3
addmargins(table(clust_k5_k7, sce_layer$subject))
# clust_k5_k7 Br5292 Br5595 Br8100 Sum
# 1 8 4 8 20
# 2 4 8 4 16
# 3 4 7 4 15
# 4 4 1 4 9
# 5 4 0 4 8
# 6 4 0 0 4
# 7 0 0 4 4
# Sum 28 20 28 76
addmargins(table(clust_k5_k7, sce_layer$layer_guess, sce_layer$subject))
# , , = Br5292
#
#