-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathbreast-1-analysis.Rmd
5423 lines (4402 loc) · 202 KB
/
breast-1-analysis.Rmd
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
---
title: "TNBC"
output: html_notebook
---
```{r}
# pacman is a "package manager" that defines the p_load function.
# p_load automatically installs a library if it is not already installed.
# This often, but doesn't always, works.
if(!require("pacman")) {
install.packages("pacman")
}
suppressPackageStartupMessages(library(pacman))
suppressPackageStartupMessages(library(Seurat))
if(!require("spacexr")) {
devtools::install_github("dmcable/spacexr", build_vignettes = FALSE)
suppressPackageStartupMessages(p_load(spacexr))
}
#install.packages("hdf5r")
#suppressPackageStartupMessages(p_load(hdf5r))
# install.packages("hdf5r", configure.args="--with-hdf5=/home/whitebr/local/hdf5/bin/h5cc")
# install.packages("https://cran.r-project.org/src/contrib/hdf5r_1.3.8.tar.gz",repos=NULL,type="source", configure.args="--with-hdf5=/home/whitebr/local/hdf5/bin/h5cc")
#install.packages("https://cran.r-project.org/src/contrib/hdf5r_1.3.8.tar.gz",repos=NULL,type="source", configure.args="--with-hdf5=/home/whitebr/local/hdf5-1.14.0/bin/h5cc")
dyn.load("/home/whitebr/local/hdf5-1.14.0/lib/libhdf5_hl.so.310")
library(hdf5r)
suppressPackageStartupMessages(p_load(ggplot2))
suppressPackageStartupMessages(p_load(edgeR))
suppressPackageStartupMessages(p_load(patchwork))
suppressPackageStartupMessages(p_load(dplyr))
suppressPackageStartupMessages(p_load(plyr))
suppressPackageStartupMessages(p_load(cowplot))
# suppressPackageStartupMessages(p_load(scales))
suppressPackageStartupMessages(p_load(reshape2))
suppressPackageStartupMessages(p_load(ggbeeswarm))
# suppressPackageStartupMessages(library(devtools))
# install_github("jokergoo/ComplexHeatmap")
suppressPackageStartupMessages(p_load(ComplexHeatmap))
# devtools::install_github("sjmgarnier/viridisLite")
suppressPackageStartupMessages(library(viridisLite))
suppressPackageStartupMessages(p_load(dendextend))
# suppressPackageStartupMessages(p_load(CCA))
# suppressPackageStartupMessages(p_load("sRDA"))
suppressPackageStartupMessages(p_load(PMA))
suppressPackageStartupMessages(p_load(ggpubr))
suppressPackageStartupMessages(p_load(rstatix))
suppressPackageStartupMessages(p_load(gtools))
suppressPackageStartupMessages(p_load(DropletUtils)) # for downsampling count matrices
suppressPackageStartupMessages(p_load(foreach))
suppressPackageStartupMessages(p_load(parallel))
suppressPackageStartupMessages(p_load(scran)) # for deconvolution-based normalization
suppressPackageStartupMessages(p_load(scuttle)) # for deconvolution-based normalization
suppressPackageStartupMessages(p_load(rjson))
##if (!require(devtools)) install.packages("devtools")
##devtools::install_github("gaospecial/ggVennDiagram")
suppressPackageStartupMessages(p_load(ggVennDiagram))
suppressPackageStartupMessages(p_load(gridExtra))
suppressPackageStartupMessages(p_load(openxlsx))
suppressPackageStartupMessages(p_load(data.table))
suppressPackageStartupMessages(p_load(biomaRt))
suppressPackageStartupMessages(library("gplots"))
suppressPackageStartupMessages(library(RColorBrewer))
suppressPackageStartupMessages(p_load(harmony))
# suppressPackageStartupMessages(p_load(spots))
# install.packages("devtools")
# devtools::install_github("stevexniu/spots")
suppressPackageStartupMessages(library(spots))
suppressPackageStartupMessages(p_load(KRLS)) # for gausskernel
suppressPackageStartupMessages(p_load(devtools))
suppressPackageStartupMessages(p_load(rhdf5))
suppressPackageStartupMessages(p_load(Rsamtools))
if(TRUE) {
if(!require(OmnipathR)) {
BiocManager::install('OmnipathR')
}
if(!require(decoupleR)) {
# install.packages("BiocManager")
# BiocManager::install("decoupleR")
devtools::install_git("https://github.com/saezlab/decoupleR")
# BiocManager::install("saezlab/decoupleR")
}
suppressPackageStartupMessages(p_load(decoupleR))
}
if(!require(dorothea)) {
BiocManager::install("dorothea")
}
suppressPackageStartupMessages(p_load(dorothea))
suppressPackageStartupMessages(p_load(dplyr))
suppressPackageStartupMessages(p_load(tibble))
suppressPackageStartupMessages(p_load(tidyr))
#require(devtools)
# #install_version("rlang", version="1.0.5", repos = "http://cran.us.r-project.org")
# remotes::install_github("satijalab/sctransform", ref="develop")
# BiocManager::install("glmGamPoi")
p_load(sctransform)
suppressPackageStartupMessages(p_load(MAST))
suppressPackageStartupMessages(p_load(transformGamPoi))
# for ridge in misty
dyn.load('/home/whitebr/local/gsl-2.5/lib/libgsl.so.23')
```
Setup a parallel execution environment.
```{r}
num.cores <- detectCores()
if(!is.na(num.cores) && (num.cores > 1)) {
suppressPackageStartupMessages(p_load("doMC"))
# num.cores <- min(num.cores,6)
cat(paste("Registering ", num.cores-1, " cores.\n", sep=""))
registerDoMC(cores=(num.cores-1))
options(mc.cores=num.cores-1)
}
```
```{r}
url <- "https://raw.githubusercontent.com/TheJacksonLaboratory/r_hne_image_utils/main/hovernet-utils.R"
source_url(url) # for extract.cell.info.from.hovernet.output, extract.hovernet.cell.info
url <- "https://raw.githubusercontent.com/TheJacksonLaboratory/cancer-cube-st-pilot-ffpe-ff/main/utils.R"
source(url)
url <- "https://raw.githubusercontent.com/TheJacksonLaboratory/cancer-cube-st-pilot-ffpe-ff/main/plotting-utils.R"
source(url)
```
```{r}
lung.init.file <- '/pod/2/activities/jaxcc/CATch-project/raw-data/visium/analysis_dir/Lung_1/lungcancer.csv'
lung.dataset.info <- read.table(lung.init.file, sep=",", header=TRUE)
dataset.info <- lung.dataset.info
output_base_dir <- '/projects/compsci/jgeorge/whitebr/cancer-cube/lung-1/'
dir.create(output_base_dir, recursive=TRUE)
datasets <- dataset.info$names
spaceranger_dirs <- dataset.info$dataset_names
names(spaceranger_dirs) <- datasets
names(datasets) <- datasets
# Define and create output results and plots directories
# base_dir_antonis<-'/projects/compsci/USERS/somara/cancer-cude/qc-pilot/'
analysis_dir <- paste0(output_base_dir, '/analysis/')
dir.create(analysis_dir, showWarnings = FALSE, recursive = TRUE)
plots_dir <- paste0(output_base_dir, '/plots/')
dir.create(plots_dir, showWarnings = FALSE, recursive = TRUE)
# Prefix for all output files (results and plots)
analysis_file_prefix <- "lung-1-"
```
```{r}
names(datasets) <- datasets
# Extract the sample name from the datasets (i.e., foo_<sample>)
patient <- dataset.info$biological_replicate
short.sample <- dataset.info$names
tissue <- dataset.info$labels
patient.df <- data.frame(sample = datasets, patient = patient, short.sample = short.sample, tissue = tissue)
rownames(patient.df) <- patient.df$sample
```
```{r}
p_load(moments)
p_load(qvalue)
tryCatch(source("https://raw.githubusercontent.com/saketkc/scRNA_NB_comparison/master/code/01_run_glm.R"),
error = function(e) print(e))
```
```{r}
source("/home/whitebr/cancer-cube-st-pilot-ffpe-ff/misty-utils.R")
```
```{r}
# net <- get_collectri(organism='human', split_complexes=FALSE)
# net <- decoupleR::get_dorothea(levels=c("A"))
```
```{r}
# Read in the filtered and unfiltered matrices and append the tissue positions information to the metadata:
filtered.objs <- create.visium.seurat.objects(spaceranger_dirs, filter.spots = TRUE)
unfiltered.objs <- create.visium.seurat.objects(spaceranger_dirs, filter.spots = FALSE)
```
```{r}
num.filtered.cells <- llply(filtered.objs, .fun = function(obj) nrow(obj[[]]))
```
```{r}
#stop("stop")
```
```{r}
indices <- 1:length(filtered.objs)
names(indices) <- names(filtered.objs)
nb.fits <-
llply(indices, .parallel = FALSE,
.fun = function(indx) {
print(names(filtered.objs)[indx])
rds.file <- paste0(analysis_dir, "nb-fits-all-", names(filtered.objs)[indx], ".rds")
if(!file.exists(rds.file)) {
fit <- suppressWarnings(GetFit(GetAssayData(filtered.objs[[indx]], assay="Spatial", slot="counts"), type="negbin"))
saveRDS(fit, file=rds.file)
}
fit <- readRDS(rds.file)
#fit$meanvarfit[, c("gene", "mean", "variance", "pval_ssrstd_quantile")]
#print(table(fit$meanvarfit$pval_ssrstd_quantile < 0.05))
#flag <- fit$meanvarfit$pval_ssrstd_quantile < 0.05
#unfit.genes <- fit$meanvarfit[flag, "gene"]
fit
})
```
```{r}
nb.ordered.fit.tbls <-
llply(nb.fits,
.fun = function(fit) {
tbl <- fit$meanvarfit
tbl <- tbl[order(tbl$pval_ssrstd_quantile, decreasing=FALSE),]
tbl
})
```
```{r}
# Perform normalization of counts using SCTransform (which fits a negative binomial to the count data)
# This adds a new assay "SCT" to the returned Seurat object. That assay has slots "counts" (the corrected counts),
# "data" (log1p(counts)), and "scale.data" (the "pearson residuals").
# These would be accessible via GetAssayData(filtered.objs[[1]], slot = "data", assay = "SCT")
# Running with default parameters (the most interesting of which look to be variable.features.n and/or variable.features.rv.th)
filtered.objs <- apply.sctransform(filtered.objs)
# glimpse(obj@[email protected]$model1)
#sct <- obj@[email protected]$model1
#data <- GetAssayData(obj, assay="Spatial", slot="counts")
#sct.feats <- [email protected]
```
```{r}
indices <- 1:nrow(patient.df)
names(indices) <- patient.df$sample
top.gs <- llply(indices,
.parallel = FALSE,
.fun = function(i) {
sample <- patient.df[i, "sample"]
short.sample <- patient.df[i, "short.sample"]
patient <- patient.df[i, "patient"]
status <- patient.df[i, "tissue"]
title <- paste0(status, " patient: ", patient, "\n(", sample, ")")
print(title)
g <- plot.top.genes(as.matrix(GetAssayData(filtered.objs[[sample]], assay="Spatial", slot="counts")))
g <- g + ggtitle(title)
})
```
```{r}
pts <- c("LM2O")
g <- plot_grid(plotlist=top.gs[patient.df$sample[patient.df$patient %in% pts]])
png(paste0(plots_dir, "top-genes-", paste(pts,collapse="-"), ".png"), width = 2*480, height=2*480)
print(g)
d <- dev.off()
```
```{r}
ffpe.sample <- "LM2O-246"
ff.sample <- "LM2O-201"
#ffpe.sample <- "EuE-31-236"
#ff.sample <- "EuE-31-164"
```
```{r}
common.genes <- intersect(rownames(GetAssayData(filtered.objs[[ff.sample]], assay="SCT", slot="counts")),
rownames(GetAssayData(filtered.objs[[ffpe.sample]], assay="SCT", slot="counts")))
ff.obj <- filtered.objs[[ff.sample]]
ff.vst.orig <- vst(GetAssayData(object = ff.obj, slot = "counts", assay = "Spatial"), vst.flavor = "v2", verbosity = 10, return_cell_attr = TRUE)
ff.obj.subset <- ff.obj[common.genes,]
ff.vst.subset <- vst(GetAssayData(object = ff.obj.subset, slot = "counts", assay = "Spatial"), vst.flavor = "v2", verbosity = 10, return_cell_attr = TRUE)
ffpe.obj <- filtered.objs[[ffpe.sample]]
ffpe.obj <- ffpe.obj[common.genes,]
ffpe.vst <- vst(GetAssayData(object = ffpe.obj, slot = "counts", assay = "Spatial"), vst.flavor = "v2", verbosity = 10, return_cell_attr = TRUE)
ff.obj.subset.same.ncells <- ff.obj[common.genes,1:length(Cells(ffpe.obj))]
ff.vst.subset.same.ncells <- vst(GetAssayData(object = ff.obj.subset.same.ncells, slot = "counts", assay = "Spatial"), vst.flavor = "v2", verbosity = 10, return_cell_attr = TRUE)
```
```{r}
#nz <- rowSums(ffpe.umi)
nz <- rowSums(ffpe.umi>10)
num.z <- rowSums(ffpe.umi==0)
num.o <- rowSums(ffpe.umi==1)
df <- data.frame(nz = nz, num.z = num.z, num.o = num.o, mn = pmin(nz, num.z))
df <- subset(df, num.o < 0.05 * num.z)
df <- df[order(df$mn, decreasing=TRUE),]
```
```{r}
gene <- "ACTB"
gene <- "EPCAM"
gene <- "Scgb1a1" # not in FFPE
gene <- "Sftpc"
gene <- "Itm2b"
gene <- "Hbb-b2"
gene <- "Hba-a2"
gene <- "Scgb3a2" # good
#genes <- c("EPCAM", "COL1A1")
gene <- "Il1b" # better
#gene <- "Pi16" # lose a low of zeros, but not normal
genes <- c(gene)
for(gene in genes) {
ff.cell_attr <- ff.vst.subset.same.ncells$cell_attr
ff.umi.all <- GetAssayData(ff.obj, assay="Spatial", slot="counts")
ff.umi <- GetAssayData(ff.obj.subset.same.ncells, assay="Spatial", slot="counts")
ff.scale_factor <- median(10^ff.cell_attr$log_umi)
ret.ff <- my_correct_counts_for_gene(ff.vst.subset.same.ncells, ff.umi, gene, cell_attr = ff.cell_attr, scale_factor = ff.scale_factor, verbosity = 0)
# fit in ret.ff$coefs[gene,] and ret.ff$theta[gene] or x$model_pars_fit[gene,]
# Substituting ffpe cell_attr, umis, and scale_factor (with ff fits)
ffpe.cell_attr <- ffpe.vst$cell_attr
ffpe.umi <- GetAssayData(ffpe.obj, assay="Spatial", slot="counts")
ffpe.scale_factor <- median(10^ffpe.cell_attr$log_umi)
ret.ffpe <- my_correct_counts_for_gene(ffpe.vst, ffpe.umi, gene, cell_attr = ffpe.cell_attr, scale_factor = ffpe.scale_factor, verbosity = 0)
# Substituting ffpe properties into ff
######################################
# Substituting ffpe cell_attr (with ff fits)
ret.ff.with.ffpe.attr <- my_correct_counts_for_gene(ff.vst.subset.same.ncells, ff.umi, gene, cell_attr = ffpe.cell_attr, scale_factor = ff.scale_factor, verbosity = 0)
# Substituting ffpe cell_attr and umis (with ff fits)
ret.ff.with.ffpe.attr.umis <- my_correct_counts_for_gene(ff.vst.subset.same.ncells, ffpe.umi, gene, cell_attr = ffpe.cell_attr, scale_factor = ff.scale_factor, verbosity = 0)
# Substituting ffpe cell_attr and scale_factor (with ff fits)
ret.ff.with.ffpe.attr.sf <- my_correct_counts_for_gene(ff.vst.subset.same.ncells, ff.umi, gene, cell_attr = ffpe.cell_attr, scale_factor = ffpe.scale_factor, verbosity = 0)
# Substituting ffpe umis (with ff fits)
ret.ff.with.ffpe.umis <- my_correct_counts_for_gene(ff.vst.subset.same.ncells, ffpe.umi, gene, cell_attr = ff.cell_attr, scale_factor = ff.scale_factor, verbosity = 0)
# Substituting ffpe umis and sf (with ff fits)
ret.ff.with.ffpe.umis.sf <- my_correct_counts_for_gene(ff.vst.subset.same.ncells, ffpe.umi, gene, cell_attr = ff.cell_attr, scale_factor = ffpe.scale_factor, verbosity = 0)
# Substituting ffpe scale factor (with ff fits)
ret.ff.with.ffpe.sf <- my_correct_counts_for_gene(ff.vst.subset.same.ncells, ff.umi, gene, cell_attr = ff.cell_attr, scale_factor = ffpe.scale_factor, verbosity = 0)
ret.ff.with.ffpe.umis.attr.sf <- my_correct_counts_for_gene(ff.vst.subset.same.ncells, ffpe.umi, gene, cell_attr = ffpe.cell_attr, scale_factor = ffpe.scale_factor, verbosity = 0)
# Substituting ff properties into ffpe
######################################
# ret.ffpe <- my_correct_counts_for_gene(ffpe.vst, ffpe.umi, gene, cell_attr = ffpe.cell_attr, scale_factor = ffpe.scale_factor, verbosity = 0)
# Substituting ff cell_attr (with ffpe fits)
ret.ffpe.with.ff.attr <- my_correct_counts_for_gene(ffpe.vst, ffpe.umi, gene, cell_attr = ff.cell_attr, scale_factor = ffpe.scale_factor, verbosity = 0)
# Substituting ff cell_attr and umis (with ffpe fits)
ret.ffpe.with.ff.attr.umis <- my_correct_counts_for_gene(ffpe.vst, ff.umi, gene, cell_attr = ff.cell_attr, scale_factor = ffpe.scale_factor, verbosity = 0)
# Substituting ff cell_attr and scale_factor (with ffpe fits)
ret.ffpe.with.ff.attr.sf <- my_correct_counts_for_gene(ffpe.vst, ffpe.umi, gene, cell_attr = ff.cell_attr, scale_factor = ff.scale_factor, verbosity = 0)
# Substituting ff umis (with ffpe fits)
ret.ffpe.with.ff.umis <- my_correct_counts_for_gene(ffpe.vst, ff.umi, gene, cell_attr = ffpe.cell_attr, scale_factor = ffpe.scale_factor, verbosity = 0)
# Substituting ff umis and sf (with ffpe fits)
ret.ffpe.with.ff.umis.sf <- my_correct_counts_for_gene(ffpe.vst, ff.umi, gene, cell_attr = ffpe.cell_attr, scale_factor = ff.scale_factor, verbosity = 0)
# Substituting ff scale factor (with ffpe fits)
ret.ffpe.with.ff.sf <- my_correct_counts_for_gene(ffpe.vst, ffpe.umi, gene, cell_attr = ffpe.cell_attr, scale_factor = ff.scale_factor, verbosity = 0)
ret.ffpe.with.ff.umis.attr.sf <- my_correct_counts_for_gene(ffpe.vst, ff.umi, gene, cell_attr = ff.cell_attr, scale_factor = ff.scale_factor, verbosity = 0)
# mu0 vs counts look very different for the two
par(mfrow=c(2,2))
hist(ff.umi[gene,], main = paste("FF raw ", gene, " nzero = ", length(which(ff.umi[gene,] <= 0))), xlab = paste0(gene, " Raw Counts"))
hist(ffpe.umi[gene,], main = paste("FFPE raw ", gene, " nzero = ", length(which(ffpe.umi[gene,] <= 0))), xlab = paste0(gene, " Raw Counts"))
hist(ret.ff$y.res[gene,], main = paste("FF vst ", gene, " nzero = ", length(which(ret.ff$y.res[gene,] <= 0))), xlab = paste0(gene, " Raw Counts"))
hist(ret.ffpe$y.res[gene,], main = paste("FFPE vst ", gene, " nzero = ", length(which(ret.ffpe$y.res[gene,] <= 0))), xlab = paste0(gene, " Raw Counts"))
par(mfrow=c(3,2))
hist(ff.umi[gene,], main = paste("FF raw ", gene, " nzero = ", length(which(ff.umi[gene,] <= 0))), xlab = paste0(gene, " Raw Counts"))
hist(ffpe.umi[gene,], main = paste("FFPE raw ", gene, " nzero = ", length(which(ffpe.umi[gene,] <= 0))), xlab = paste0(gene, " Raw Counts"))
hist(ret.ff$y.res[gene,], main = paste("FF vst ", gene, " nzero = ", length(which(ret.ff$y.res[gene,] <= 0))), xlab = paste0(gene, " Raw Counts"))
hist(ret.ffpe$y.res[gene,], main = paste("FFPE vst ", gene, " nzero = ", length(which(ret.ffpe$y.res[gene,] <= 0))), xlab = paste0(gene, " Raw Counts"))
my_hist(ff.umi[gene,]/ff.umi["Actb",], xlims=c(0,5), main = paste("FF raw ", gene, " / Actb nzero = ", length(which((ff.umi[gene,]/ff.umi["Gapdh",])==0))))
my_hist(ffpe.umi[gene,]/ffpe.umi["Actb",], xlims=c(0,0.5), main = paste("FFPE raw ", gene, " / Actb nzero = ", length(which((ffpe.umi[gene,]/ffpe.umi["Actb",])==0))))
m <- merge(ff.vst.subset$model_pars_fit, ffpe.vst$model_pars_fit, by = "row.names")
# Intercepts are very similar
plot(m[, "(Intercept).x"], m[, "(Intercept).y"])
# Thetas are not
plot(m$theta.x, m$theta.y)
ff.cc <- correct_counts(ff.vst.orig, GetAssayData(ff.obj, assay="Spatial", slot="counts"),verbosity=20)
ff.cc.subset <- correct_counts(ff.vst.subset, GetAssayData(ff.obj.subset, assay="Spatial", slot="counts"),verbosity=20)
ffpe.cc <- correct_counts(ffpe.vst, GetAssayData(ffpe.obj, assay="Spatial", slot="counts"),verbosity=20)
# WORKING
# Observation: EPCAM has zero raw counts in FFPE, but few / no zero SCT-transformed counts in FFPE
# Why?
# Confirm for x_c = 0, y_c > 0 if n0 > nc
# Other observations:
# 1. Tot counts are normal in FF, but skewed in FFPE
# 2. mu_0 is exactly proportional to total counts. This is following Lause.
# 3. pearson residuals distibutions look very different
# 4. Given how simple the fit is, I'm not sure it makes sense to switch out the counts and cell_attrs as I do below.
# 5. mu_0 > 0 because mu_0 = nc * e^beta_0 and both nc and e^beta_0 are > 0
# 6. Very strong correlation in ffpe of gene UMI with total counts (at least for epcam), less true for ff
flag <- (ret.ff.with.ffpe.attr.umis$y[gene,]==0)
hist(ret.ff.with.ffpe.attr.umis$y.res[gene,flag])
plot(ret.ff$mu_0[gene,], colSums(ff.umi))
plot(ret.ff$mu_0[gene,], ff.umi[gene,])
# WORKING: when we use the same umis and the same cell_attrs, but use different fits,
# the results look very similar (ie hist(y.res.ff))
# Confirm this.
# What can we change to break this? umis? cell_attr? scale_factor?
# In the other direction, confirm that using the same fits, but different counts,
# the results look very different.
ffpe.zero.flag <- ret.ffpe$y.res[gene,] <= 0
# mu0 vs counts look very different for the two
par(mfrow=c(1,2))
plot(ff.umi[gene,], ret.ff$mu_0[gene,])
abline(a=0,b=1)
plot(ffpe.umi[gene,], ret.ffpe$mu_0[gene,])
abline(a=0,b=1)
text(ffpe.umi[gene,ffpe.zero.flag], ret.ffpe$mu_0[gene,ffpe.zero.flag], labels="X")
# mu0 vs counts look very different for the two
par(mfrow=c(1,2))
plot(ff.umi[gene,], ret.ff$mu_0[gene,], xlim=c(0,10))
abline(a=0,b=1)
plot(ffpe.umi[gene,], ret.ffpe$mu_0[gene,], xlim=c(0,10))
abline(a=0,b=1)
text(ffpe.umi[gene,ffpe.zero.flag], ret.ffpe$mu_0[gene,ffpe.zero.flag], labels="X")
par(mfrow=c(1,2))
plot(ff.umi[gene,], ff.umi[gene,] - ret.ff$mu_0[gene,], xlim = c(0,10))
plot(ffpe.umi[gene,], ffpe.umi[gene,] - ret.ffpe$mu_0[gene,], xlim = c(0,10))
text(ffpe.umi[gene,ffpe.zero.flag], ffpe.umi[gene,ffpe.zero.flag] - ret.ffpe$mu_0[gene,ffpe.zero.flag], labels="X")
par(mfrow=c(1,2))
plot(ff.umi[gene,], ret.ff$pearson_residual[gene,], xlim = c(0,10))
plot(ffpe.umi[gene,], ret.ffpe$pearson_residual[gene,], xlim = c(0,10))
text(ffpe.umi[gene,ffpe.zero.flag], ret.ffpe$pearson_residual[gene,ffpe.zero.flag], labels="X")
par(mfrow=c(1,2))
plot(ff.umi[gene,], sqrt(ret.ff$variance_0[gene,]), xlim = c(0,10))
plot(ffpe.umi[gene,], sqrt(ret.ffpe$variance_0[gene,]), xlim = c(0,10))
text(ffpe.umi[gene,ffpe.zero.flag], sqrt(ret.ffpe$variance_0[gene,ffpe.zero.flag]), labels="X")
par(mfrow=c(1,2))
# Original ff results look skewed
my_hist(ret.ff$y.res[gene,], xlims = c(-Inf,30), main = "FF orig")
# ff fit with ffpe umis and cell attr looks normal
my_hist(ret.ffpe$y.res[gene,], xlims = c(0,30), main = "FFPE orig")
# Theta for ffpe is _much_ bigger:
print(c(ret.ffpe$theta[gene], ret.ff$theta[gene]))
# Should imply variances are smaller (but also a function of mu)
# Variance of FFPE is right-skewed, because mu_0 is right-skewed, because tot counts are right skewed
par(mfrow=c(1,2))
my_hist(ret.ff$variance_0[gene,], xlims = c(-Inf,30), main = "FF orig")
my_hist(ret.ffpe$variance_0[gene,], xlims = c(0,30), main = "FFPE orig")
par(mfrow=c(1,2))
hist(ret.ff$pearson_residual[gene,])
hist(ret.ffpe$pearson_residual[gene,])
par(mfrow=c(1,2))
plot(sqrt(ret.ff$variance_0[gene,]), sqrt(ret.ff$mu_0[gene,]))
abline(a=0,b=1)
plot(sqrt(ret.ffpe$variance_0[gene,]), sqrt(ret.ffpe$mu_0[gene,]))
abline(a=0,b=1)
# Pearson residual and pearson * sqrt(variance) are skeweed in ff
par(mfrow=c(1,2))
hist((ret.ff$pearson_residual[gene,] * sqrt(ret.ff$variance[gene,])))
hist((ret.ffpe$pearson_residual[gene,] * sqrt(ret.ffpe$variance[gene,])))
par(mfrow=c(1,2))
hist((ret.ff$pearson_residual[gene,] * sqrt(ret.ff$variance[gene,])) + ret.ff$mu[gene,])
hist((ret.ffpe$pearson_residual[gene,] * sqrt(ret.ffpe$variance[gene,])) + ret.ffpe$mu[gene,])
par(mfrow=c(1,2))
plot(colSums(ff.umi), ret.ff$pearson_residual[gene,])
plot(colSums(ffpe.umi), ret.ffpe$pearson_residual[gene,])
# mu and variance are the same for all cells (for all genes), though the values differ by gene.
# This is again a consequence of Lause, with a fixed umi count n0
par(mfrow=c(1,2))
hist(ret.ff$variance[gene,], main = "FF orig")
hist(ret.ffpe$variance[gene,], main = "FFPE orig")
par(mfrow=c(1,2))
hist(ret.ff$mu[gene,], main = "FF orig")
hist(ret.ffpe$mu[gene,], main = "FFPE orig")
par(mfrow=c(1,2))
my_hist(ret.ff$mu_0[gene,], xlims = c(-Inf,30), main = "FF orig")
my_hist(ret.ffpe$mu_0[gene,], xlims = c(0,30), main = "FFPE orig")
par(mfrow=c(1,2))
hist((colSums(ff.umi)))
hist((colSums(ffpe.umi)))
par(mfrow=c(1,2))
hist(ret.ff$variance[gene,], main = "FF orig")
hist(ret.ffpe$variance[gene,], main = "FFPE orig")
# Plot ff with ffpe subs
par(mfrow=c(3,3))
# Original ff results look skewed
my_hist(ret.ff$y.res[gene,], xlims = c(-Inf,30), main = "FF orig")
# ff fit with ffpe umis and cell attr looks normal
my_hist(ret.ff.with.ffpe.umis.attr.sf$y.res[gene,], xlims = c(0,30), main = "FF sub umis, attr, sf")
my_hist(ret.ff.with.ffpe.umis$y.res[gene,], xlims = c(-Inf,30), main = "FF sub umis")
my_hist(ret.ff.with.ffpe.attr$y.res[gene,], xlims = c(-Inf,30), main = "FF sub attr")
my_hist(ret.ff.with.ffpe.sf$y.res[gene,], xlims = c(-Inf,30), main = "FF sub sf")
my_hist(ret.ff.with.ffpe.attr.umis$y.res[gene,], xlims = c(-Inf,30), main = "FF sub umis, attr")
my_hist(ret.ff.with.ffpe.attr.sf$y.res[gene,], xlims = c(-Inf,30), main = "FF sub attr, sf")
my_hist(ret.ff.with.ffpe.umis.sf$y.res[gene,], xlims = c(-Inf,30), main = "FF sub umis, sf")
# Plot ffpe with ff subs
par(mfrow=c(3,3))
my_hist(ret.ffpe$y.res[gene,], xlims = c(-Inf,30), main = "FFPE orig")
my_hist(ret.ffpe.with.ff.umis.attr.sf$y.res[gene,], xlims = c(0,30), main = "FFPE sub umis, attr, sf")
my_hist(ret.ffpe.with.ff.umis$y.res[gene,], xlims = c(-Inf,30), main = "FFPE sub umis")
my_hist(ret.ffpe.with.ff.attr$y.res[gene,], xlims = c(-Inf,30), main = "FFPE sub attr")
my_hist(ret.ffpe.with.ff.sf$y.res[gene,], xlims = c(-Inf,30), main = "FFPE sub sf")
my_hist(ret.ffpe.with.ff.attr.umis$y.res[gene,], xlims = c(-Inf,30), main = "FFPE sub umis, attr")
my_hist(ret.ffpe.with.ff.attr.sf$y.res[gene,], xlims = c(-Inf,30), main = "FFPE sub attr, sf")
my_hist(ret.ffpe.with.ff.umis.sf$y.res[gene,], xlims = c(-Inf,30), main = "FFPE sub umis, sf")
compare_vsts(ret.ff, "ff", ret.ff.with.ffpe.attr.umis, "ff.with.ffpe.attr.umis", gene)
par(mfrow=c(1,2))
hist(ret.ff$pearson_residual[gene,])
hist(ret.ff.with.ffpe.attr.umis$pearson_residual[gene,])
# Plot UMIs for gene vs colsums
par(mfrow=c(1,2))
plot(ff.umi[gene,], colSums(ff.umi))
plot(ffpe.umi[gene,], colSums(ffpe.umi))
g <- ggplot_smooth_scatter(data.frame(x=ff.umi[gene,], y=colSums(ff.umi)), aes(x=x,y=y))
# Plot UMIs for gene vs colsums (limit x axis)
par(mfrow=c(1,2))
plot(ff.umi[gene,], colSums(ff.umi), xlim=c(0,30))
plot(ffpe.umi[gene,], colSums(ffpe.umi), xlim=c(0,30))
par(mfrow=c(1,2))
plot(colSums(ff.umi), ff.umi[gene,])
plot(colSums(ffpe.umi), ffpe.umi[gene,])
par(mfrow=c(1,2))
plot(log10(colSums(ff.umi)), ff.umi[gene,])
plot(log10(colSums(ffpe.umi)), ffpe.umi[gene,])
par(mfrow=c(1,2))
hist(colSums(ffpe.umi))
hist(colSums(ff.umi))
par(mfrow=c(1,2))
my_hist(log(ret.ff.with.ffpe.attr.umis$mu_0[gene,]))
my_hist(log(ret.ff$mu_0[gene,]))
par(mfrow=c(1,2))
plot(colSums(ret.ff.with.ffpe.attr.umis$y), log(ret.ff.with.ffpe.attr.umis$mu_0[gene,]))
plot(colSums(ret.ff$y), log(ret.ff$mu_0[gene,]))
par(mfrow=c(1,2))
y.min.1 <- min(log(ret.ff.with.ffpe.attr.umis$mu_0[gene,]))
y.min.2 <- min(log(ret.ff$mu_0[gene,]))
y.max <- max(log(ret.ff$mu_0[gene,]))
plot(colSums(ret.ff.with.ffpe.attr.umis$y), log(ret.ff.with.ffpe.attr.umis$mu_0[gene,]), ylim=c(y.min.1, y.max), xlim=c(0,1000))
plot(colSums(ret.ff$y), log(ret.ff$mu_0[gene,]), ylim=c(y.min.2, y.max), xlim=c(0,1000))
}
```
```{r}
stop("stop")
```
```{r}
cpms <-
llply(filtered.objs,
.fun = function(obj) {
mat <- GetAssayData(obj, assay="Spatial", slot="counts")
norm.mat <- edgeR::cpm(as.matrix(mat), log=FALSE)
norm.mat
})
```
```{r}
l.cpms <-
llply(filtered.objs,
.fun = function(obj) {
mat <- GetAssayData(obj, assay="Spatial", slot="counts")
norm.mat <- edgeR::cpm(as.matrix(mat), log=TRUE)
norm.mat
})
```
```{r}
# MISTy analysis
p_load(mistyR)
p_load(future)
# data manipulation
p_load(dplyr)
p_load(purrr)
p_load(distances)
plan(multisession)
p_load(tidyverse)
setwd("/home/whitebr/local/gsl-2.5/lib/")
```
```{r}
net <- get_progeny(organism = 'human', top = 100)
common.genes <- Reduce(intersect, lapply(filtered.objs, function(obj) rownames(obj)))
net.sub <- subset(net, target %in% common.genes)
# Get ligands from omnipath
icn <- import_intercell_network()
icn.ligand <- subset(icn, category_intercell_source == "ligand")
# Find the highest confident 100 ligands
hc.ligands <-
ddply(as.data.frame(icn.ligand[, c("source_genesymbol", "consensus_score_intercell_source")]),
.variables = c("source_genesymbol"),
.fun = function(df) {
data.frame(score = max(df$consensus_score_intercell_source))
})
hc.ligands <- hc.ligands[order(hc.ligands$score, decreasing=TRUE),]
colnames(hc.ligands) <- c("gene", "score")
hc.ligands <- subset(hc.ligands, gene %in% common.genes)
net.sub.no.ligands <- subset(net.sub, !(target %in% hc.ligands$gene))
avg.max.lig.expr <-
ldply(filtered.objs,
.fun = function(obj) {
vec <- apply(GetAssayData(obj, assay="Spatial", slot="counts"),1,max)[hc.ligands$gene]
})
colnames(avg.max.lig.expr)[1] <- "sample"
rownames(avg.max.lig.expr) <- avg.max.lig.expr$sample
avg.max.lig.expr <- avg.max.lig.expr[, !(colnames(avg.max.lig.expr) %in% c("sample"))]
sms <- colSums(avg.max.lig.expr)
sms <- sms[order(sms, decreasing=TRUE)]
top.ligands <- unique(c("MFGE8", names(sms)[1:35]))
```
```{r}
all.sample.acts <-
llply(l.cpms,
.fun = function(expr) {
sample_acts <- run_wmean(mat=expr, net=net.sub.no.ligands, .source='source', .target='target',
.mor='weight', times = 2, minsize = 5)
})
```
```{r}
stop("misty")
```
```{r}
sample.name <- names(filtered.objs)[1]
for(sample.name in names(filtered.objs)) {
print(sample.name)
rctd.df <- format.rctd.output_(rctds[[sample.name]])
rctd.df <- rctd.df[, !(colnames(rctd.df) %in% c("x","y"))]
sample_acts <- subset(all.sample.acts[[sample.name]], statistic == "wmean")[, c("source", "condition", "score")]
act.mat <- dcast(sample_acts, condition ~ source)
rownames(act.mat) <- act.mat$condition
act.mat <- act.mat[, !(colnames(act.mat) %in% c("condition"))]
colnames(act.mat) <- make.names(colnames(act.mat))
expr <- l.cpms[[sample.name]]
obj <- filtered.objs[[sample.name]]
common.spots <- intersect(rownames(act.mat), colnames(expr))
act.mat <- act.mat[common.spots,]
rctd.df <- rctd.df[common.spots,]
obj <- obj[, common.spots]
expr <- expr[, common.spots]
exclude <- apply(rctd.df, 1, function(row) any(is.na(row)))
act.mat <- act.mat[!exclude,]
rctd.df <- rctd.df[!exclude,]
expr <- expr[, !exclude]
obj <- obj[, !exclude]
pos <- obj[[]][,c("row","col")]
obj <- add.metadata.to.seurat.obj(obj, act.mat)
#tmp <- rbind(expr[rownames(expr) %in% top.ligands,], t(rctd.df))
tmp <- expr[rownames(expr) %in% top.ligands,]
ligand.expr <- tmp %>% t() %>% as_tibble(rownames = NA)
act.expr <- act.mat %>% as_tibble(rownames = NA)
misty.views <- create_initial_view(act.expr)
# paraview for pathways
misty.views <- misty.views %>% add_paraview(pos, l = 9, zoi = 1)
names(misty.views)[length(misty.views)] <- "paraview.path_9"
misty.views[[length(misty.views)]]$abbrev <- "para.path_9"
# paraview for ligands (and cell types)
# misty.views <- misty.views %>% add_views(create_view("paraview.lig_9", ligand.expr, "para.lig_9"))
dummy.views <- create_initial_view(ligand.expr)
dummy.views <- dummy.views %>% add_paraview(pos, l = 9, zoi = 0)
data.red <- dummy.views[["paraview.9"]]$data
misty.views <- misty.views %>% add_views(create_view("paraview.lig_9", data.red, "para.lig_9"))
#dummy.views <- create_initial_view(cell.type.freq)
#dummy.views <- dummy.views %>% add_paraview(pos, l = 9, zoi = 0)
#data.red <- dummy.views[["paraview.9"]]$data
#misty.views <- misty.views %>% add_views(create_view("paraview.ct_9", data.red, "para.ct_9"))
res <- run_misty(misty.views, results.folder = paste0(analysis_dir, "misty-path-lig-indep-", sample.name))
#%>% collect_results()
}
```
```{r}
ffpe.sample <- "EuE-31-235"
ff.sample <- "EuE-31-165"
#ffpe.sample <- "EuE-31-236"
#ff.sample <- "EuE-31-164"
```
```{r}
ffpe.obj <- filtered.objs[[ffpe.sample]]
ff.obj <- filtered.objs[[ff.sample]]
ffpe.sct <- ffpe.obj@[email protected]$model1
ffpe.data <- GetAssayData(ffpe.obj, assay="Spatial", slot="counts")
ffpe.sct.feats <- [email protected]
ff.sct <- ff.obj@[email protected]$model1
ff.data <- GetAssayData(ff.obj, assay="Spatial", slot="counts")
ff.sct.feats <- [email protected]
genes <- rownames(ffpe.data)
names(genes) <- genes
ffpe.cs <- colSums(ffpe.data)
cs.cors <- llply(genes[1:10], .fun = function(gene) cor(ffpe.data[gene,], ffpe.cs))
cs.cors <- cs.cors[order(cs.cors, decreasing=TRUE),]
cs.cors <- cor(t(as.matrix(ff.data)),colSums(ff.data))
cs.cors <- cs.cors[order(cs.cors[,1], decreasing=TRUE),]
# plot(log10(sct.feats$gmean), sct.feats[, "step1_(Intercept)"])
plot(log10(ff.sct.feats$gmean), log10(ff.sct.feats$variance))
plot(log10(ffpe.sct.feats$gmean), ffpe.sct.feats$residual_variance)
hist(colSums(ffpe.data))
hist(colSums(ff.data))
plot(density(log10(ffpe.sct.feats$gmean)))
lines(density(log10(ff.sct.feats$gmean)))
```
```{r}
res <- collect_results(paste0(analysis_dir, "misty-cell-path-lig-indep-", ff.sample))
```
```{r}
res %>% plot_improvement_stats("gain.R2")
res %>% plot_view_contributions()
res %>% plot_interaction_heatmap(view = "intra")
res %>% plot_interaction_heatmap(view = "para.lig_9")
res %>% plot_interaction_heatmap(view = "para.path_9")
res$importances.aggregated %>% filter(view == "para.lig_9", Target == "Hypoxia") %>% arrange(-Importance)
```
```{r}
my_correct_counts_for_gene <-
function (x, umi, gene, cell_attr = x$cell_attr, scale_factor = NA,
verbosity = 2, verbose = NULL, show_progress = NULL)
{
if (!is.null(verbose)) {
warning("The 'verbose' argument is deprecated as of v0.3. Use 'verbosity' instead. (in sctransform::vst)",
immediate. = TRUE, call. = FALSE)
verbosity <- as.numeric(verbose)
}
if (!is.null(show_progress)) {
warning("The 'show_progress' argument is deprecated as of v0.3. Use 'verbosity' instead. (in sctransform::vst)",
immediate. = TRUE, call. = FALSE)
if (show_progress) {
verbosity <- 2
}
else {
verbosity <- min(verbosity, 1)
}
}
regressor_data_orig <- model.matrix(as.formula(gsub("^y",
"", x$model_str)), cell_attr)
cell_attr[, x$arguments$latent_var] <- apply(cell_attr[,
x$arguments$latent_var, drop = FALSE], 2, function(x) rep(median(x),
length(x)))
if (!is.na(scale_factor) && !is.numeric(scale_factor)) {
stop("`scale_factor` should be numeric")
}
if (!is.na(scale_factor)) {
if (verbosity > 0) {
message(paste("Setting log_umi for correcting counts to",
scale_factor))
}
cell_attr[, "log_umi"] <- log10(scale_factor)
}
regressor_data <- model.matrix(as.formula(gsub("^y", "",
x$model_str)), cell_attr)
genes <- rownames(umi)[rownames(umi) %in% rownames(x$model_pars_fit)]
bin_size <- x$arguments$bin_size
bin_ind <- ceiling(x = 1:length(x = genes)/bin_size)
max_bin <- max(bin_ind)
if (verbosity > 0) {
message("Computing corrected UMI count matrix")
}
if (verbosity > 1) {
pb <- txtProgressBar(min = 0, max = max_bin, style = 3)
}
corrected_data <- list()
i <- bin_ind[which(genes==gene)]
# for (i in 1:max_bin) {
genes_bin <- genes[bin_ind == i]
coefs <- x$model_pars_fit[genes_bin, -1, drop = FALSE]
theta <- x$model_pars_fit[genes_bin, 1]
mu_0 <- exp(tcrossprod(coefs, regressor_data_orig))
variance_0 <- mu_0 + mu_0^2/theta
y <- as.matrix(umi[genes_bin, , drop = FALSE])
pearson_residual <- (y - mu_0)/sqrt(variance_0)
mu <- exp(tcrossprod(coefs, regressor_data))
variance <- mu + mu^2/theta
y.res <- mu + pearson_residual * sqrt(variance)
# y.res <- round(y.res, 0)
# y.res[y.res < 0] <- 0
# corrected_data[[length(corrected_data) + 1]] <- make.sparse(mat = y.res)
if (verbosity > 1) {
setTxtProgressBar(pb, i)
}
return(list(coefs = coefs, theta = theta, mu_0 = mu_0, variance_0 = variance_0, y = y, pearson_residual = pearson_residual, mu = mu, variance = variance, y.res = y.res, regressor_data_orig = regressor_data_orig, regressor_data = regressor_data))
# }
if (verbosity > 1) {
close(pb)
}
corrected_data <- do.call(what = rbind, args = corrected_data)
return(corrected_data)
}
```
```{r}
my_hist <- function(x, xlims = c(0,30), ...) {
x <- x[x >= xlims[1]]
x <- x[x <= xlims[2]]
hist(x, ...)
}
```
```{r}
compare_vsts <- function(vst1, vst1.name, vst2, vst2.name, gene) {
vars <- c("coefs", "theta", "mu_0", "variance_0", "pearson_residual", "mu", "variance")
for(var in vars) {
cat(paste0(var, "\n"))
cat(paste0(vst1.name, "\n"))
if(is.null(dim(vst1[[var]]))) {
print(vst1[[var]][gene])
} else {
print(head(vst1[[var]][gene,]))
}
cat(paste0(vst2.name, "\n"))
if(is.null(dim(vst2[[var]]))) {
print(vst2[[var]][gene])
} else {
print(head(vst2[[var]][gene,]))
}
cat("\n")
}
}
```
```{r}
gene <- "ACTB"
gene <- "EPCAM"
for(gene in c("EPCAM", "COL1A1")) {
common.genes <- intersect(rownames(GetAssayData(filtered.objs[[ff.sample]], assay="SCT", slot="counts")),
rownames(GetAssayData(filtered.objs[[ffpe.sample]], assay="SCT", slot="counts")))
ff.obj <- filtered.objs[[ff.sample]]
ff.vst.orig <- vst(GetAssayData(object = ff.obj, slot = "counts", assay = "Spatial"), vst.flavor = "v2", verbosity = 10, return_cell_attr = TRUE)
ff.obj.subset <- ff.obj[common.genes,]
ff.vst.subset <- vst(GetAssayData(object = ff.obj.subset, slot = "counts", assay = "Spatial"), vst.flavor = "v2", verbosity = 10, return_cell_attr = TRUE)
ffpe.obj <- filtered.objs[[ffpe.sample]]
ffpe.obj <- ffpe.obj[common.genes,]
ffpe.vst <- vst(GetAssayData(object = ffpe.obj, slot = "counts", assay = "Spatial"), vst.flavor = "v2", verbosity = 10, return_cell_attr = TRUE)
ff.obj.subset.same.ncells <- ff.obj[common.genes,1:length(Cells(ffpe.obj))]
ff.vst.subset.same.ncells <- vst(GetAssayData(object = ff.obj.subset.same.ncells, slot = "counts", assay = "Spatial"), vst.flavor = "v2", verbosity = 10, return_cell_attr = TRUE)
m <- merge(ff.vst.subset$model_pars_fit, ffpe.vst$model_pars_fit, by = "row.names")
# Intercepts are very similar
plot(m[, "(Intercept).x"], m[, "(Intercept).y"])
# Thetas are not
plot(m$theta.x, m$theta.y)
ff.cc <- correct_counts(ff.vst.orig, GetAssayData(ff.obj, assay="Spatial", slot="counts"),verbosity=20)
ff.cc.subset <- correct_counts(ff.vst.subset, GetAssayData(ff.obj.subset, assay="Spatial", slot="counts"),verbosity=20)
ffpe.cc <- correct_counts(ffpe.vst, GetAssayData(ffpe.obj, assay="Spatial", slot="counts"),verbosity=20)
# WORKING
# Observation: EPCAM has zero raw counts in FFPE, but few / no zero SCT-transformed counts in FFPE
# Why?
# Confirm for x_c = 0, y_c > 0 if n0 > nc
# Other observations:
# 1. Tot counts are normal in FF, but skewed in FFPE
# 2. mu_0 is exactly proportional to total counts. This is following Lause.
# 3. pearson residuals distibutions look very different
# 4. Given how simple the fit is, I'm not sure it makes sense to switch out the counts and cell_attrs as I do below.