-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathMap2.2_MySeuratPipeline.R
329 lines (263 loc) · 12.7 KB
/
Map2.2_MySeuratPipeline.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
require(methods)
script_dir = "/cluster/home/tandrews/scripts/LiverMap2.0"
auto_anno_dir = "/cluster/projects/macparland/TA/AutoAnnotation"
## Auto Annotation Stuff##
source(paste(script_dir, "Colour_Scheme.R", sep="/"))
simplify_annotations <- function(annotations) {
simplified <- as.character(annotations)
simplified[simplified %in% c(
"AntibodysecretingBcells",
"MatureBcells", "B_cells", "any_B_cell", "Anti_B")] <- "Bcells"
simplified[simplified %in% c(
"CD3abTcells", "gdTcells1", "gdTcells2", "gd_T_1", "gd_T_2", "CD3_T")] <- "Tcells"
simplified[simplified %in% c(
"PericentralHep", "UnidentifiedHep", "PeriportalHep",
"interzonalHep", "Hep", "PortalHep2", "PortaHep1", "PortHep", "CentralHep1", "CentralHep")] <- "Hepatocyte"
return(map_cell_types(simplified));
}
do_annotation <- function(myseur) {
myseur <- run_scmap_seurat(myseur, scmap_ref=map1_ref);
marker_anno <- Use_markers_for_anno(myseur@assays$RNA@data, myseur$seurat_clusters);
general_labs <- simplify_annotations([email protected]$scmap_cell_anno)
general_labs2 <- simplify_annotations([email protected]$scmap_cluster_anno)
general_labs[general_labs != general_labs2] <- "ambiguous"
general_labs[general_labs == "unassigned"] <- "ambiguous"
[email protected]$general_labs <- general_labs;
inconsistent <- [email protected]$scmap_cell_anno != [email protected]$scmap_cluster_anno;
[email protected]$consistent_labs <- as.character([email protected]$scmap_cell_anno);
[email protected]$consistent_labs[inconsistent] <- as.character(general_labs[inconsistent]);
[email protected]$marker_labs <- as.character(marker_anno$cell_assign)
[email protected]$marker_general_labs <- simplify_annotations(as.character(marker_anno$cell_assign))
return(myseur)
}
source(paste(script_dir, "Setup_autoannotation.R", sep="/"))
args <- as.character(commandArgs(trailingOnly=TRUE))
Params <- read.table(paste(script_dir, "LiverMap_SampleProcessingParams.csv", sep="/"), sep=",", header=T)
if (grepl("rds", args[1])) {
mt_filter <- 50
ng_filter <- 100
nc_filter <- 10
nkNN <- 20
npcs <- 20
nhvg <- 2000
this_seed <- 101
max_cells <- 20000
file = args[1]
f <- unlist(strsplit(file, "/"))
f <- f[length(f)]
f <- unlist(strsplit(f, "[_\\.]"))
name <- paste(f[1:(length(f)-3)], collapse="_");
rds <- file
out_tag = paste(c(f[(length(f)-2):(length(f)-1)], "SeurObj"), collapse="_")
folder <- as.character(Params[Params$Name==name,"Directory_UHN"])
} else {
i <- as.numeric(args[1])
mt_filter <- Params[i,"MTfilter"]
ng_filter <- Params[i,"nGenefilter"]
nc_filter <- Params[i,"nCellfilter"]
nkNN <- Params[i, "kNN"]
npcs <- Params[i, "nPCs"]
nhvg <- Params[i, "nHGV"]
this_seed <- Params[i, "Seed"]
max_cells <- 20000
name <- as.character(Params[i, "Name"])
folder <- as.character(Params[i, "Directory_UHN"])
rds <- as.character(Params[i, "RDS_file"])
out_tag = "Anno_SeurObj"
}
print(paste(name,"_", out_tag, ".rds", sep=""))
if (!file.exists( paste(name,"_", out_tag, ".rds", sep=""))) {
set.seed(this_seed)
#res <- 5
#res <- 3
# chanaged on 13-07-2020
res <- 1
print("Read Data")
require(dplyr)
require(Seurat)
require(Matrix)
if (file.exists(rds)) {
mydata <- readRDS(rds)
} else {
my_10xfolder <- paste(folder, "filtered_gene_bc_matrices/GRCh38", sep="/")
# Read the data
mydata <- Seurat::Read10X(data.dir = paste(folder, "filtered_gene_bc_matrices/GRCh38", sep="/"))
print(paste("Stats out:", dim(mydata), "input dimensions of", name))
# Enforce a maximum number of cells captured based on expected number.
mydata <- mydata[,Matrix::colSums(mydata) > ng_filter]
if (ncol(mydata) > 20000) {
ng_detect <- Matrix::colSums(mydata);
q_thresh <- quantile(ng_detect, 1-max_cells/length(ng_detect))
mydata <- mydata[,ng_detect > q_thresh]
}
}
myseur <- Seurat::CreateSeuratObject(counts = mydata, project = name, min.cells = nc_filter, min.features = ng_filter) ### <- ERROR! gene & cell filter are backwards. - Fixed on 20/04/2020
# Mitochondrial filter
myseur[["percent.mt"]] <- Seurat::PercentageFeatureSet(myseur, pattern = "^MT-")
myseur <- subset(myseur, subset = nFeature_RNA > ng_filter & percent.mt < mt_filter)
print(paste("Stats out:", dim(myseur), "seurat dimensions of", name))
# Add metadata
[email protected]$cell_barcode <- colnames(myseur)
[email protected]$donor <- rep(name, ncol(myseur));
[email protected]$cell_ID <- paste([email protected]$donor, [email protected]$cell_barcode, sep="_");
#add corrected table
if (grepl("emptyDrops", rds)) {
corrected_rds <- gsub("table", "correctedtable", rds)
corrected_mat <- readRDS(corrected_rds)
corrected_mat <- corrected_mat[,match(colnames(myseur), colnames(corrected_mat))]
corrected_mat <- corrected_mat[match(rownames(myseur), rownames(corrected_mat)),]
myseur@assays$My_Corrected <- corrected_mat;
[email protected]$corrected_percent.mt <- Matrix::colSums(corrected_mat[
grepl("MT-",rownames(corrected_mat)),])/
Matrix::colSums(corrected_mat)
}
print("Seurat Pipeline")
# sctransform normalization
#require("sctransform")
#norm <- sctransform::vst(Matrix(myseur@assays$RNA@counts), res_clip_range=c(-Inf, Inf), method="nb_fast");
#myseur@assays$RNA@data <- norm$y;
myseur <- Seurat::NormalizeData(myseur);
# Scale
myseur <- Seurat::ScaleData(myseur);
# HVG genes (n=2000)
myseur <- Seurat::FindVariableFeatures(myseur, selection.method = "vst", nfeatures = 2000)
# PCA
myseur <- Seurat::RunPCA(myseur, features = VariableFeatures(object = myseur))
ElbowPlot(myseur)
# Clustering
myseur <- Seurat::FindNeighbors(myseur, dims = 1:npcs)
myseur <- Seurat::FindClusters(myseur, resolution = res, k.param=nkNN)
# Visualization with TSNE & UMAP
myseur <- Seurat::RunTSNE(myseur, dims = 1:npcs)
myseur <- Seurat::RunUMAP(myseur, dims = 1:npcs, parallel=FALSE)
#Cell-cycle
s.genes <- cc.genes$s.genes
g2m.genes <- cc.genes$g2m.genes
myseur <- Seurat::CellCycleScoring(myseur, s.features = s.genes, g2m.features=g2m.genes, set.ident=TRUE)
# AutoAnnotation with scmap
myseur <- do_annotation(myseur)
saveRDS(myseur, paste(name,"_",out_tag, ".rds", sep=""));
} else {
myseur <- readRDS( paste(name,"_",out_tag, ".rds", sep=""));
}
print(paste("Stats out:", length(unique([email protected]$seurat_clusters)), "seurat clusters in", name))
##### Make plots ####
agg_coord_by_cluster <- function(coords, clusters) {
x <- split(seq(nrow(coords)), clusters)
result <- sapply(x, function(a) apply(coords[a,],2,median))
return(result)
}
umap_lab_pos <- agg_coord_by_cluster(myseur@[email protected], [email protected]$seurat_clusters)
require("ggplot2")
require("Seurat")
# UMAP + Ref scmap anno
new_colour_scheme <- Cell_type_colours[order(Cell_type_colours[,1]),]
[email protected]$consistent_labs <- map_cell_types([email protected]$consistent_labs)
new_colour_scheme <- new_colour_scheme[new_colour_scheme[,1] %in% [email protected]$consistent_labs,]
png(paste(name, out_tag, "_refanno_umap.png", sep="_"), width=6, height=6, units="in", res=100)
DimPlot(myseur, reduction="umap", group.by="consistent_labs", pt.size=.1)+scale_color_manual(values=new_colour_scheme[,2])+annotate("text", x=umap_lab_pos[1,], y=umap_lab_pos[2,], label=colnames(umap_lab_pos), colour="grey35")
dev.off()
# UMAP + Marker scmap anno
new_colour_scheme <- Cell_type_colours[order(Cell_type_colours[,1]),]
[email protected]$marker_labs <- map_cell_types([email protected]$marker_labs)
new_colour_scheme <- new_colour_scheme[new_colour_scheme[,1] %in% [email protected]$marker_labs,]
png(paste(name, out_tag, "_markanno_umap.png", sep="_"), width=6, height=6, units="in", res=100)
DimPlot(myseur, reduction="umap", group.by="marker_labs", pt.size=.1)+scale_color_manual(values=new_colour_scheme[,2])+annotate("text", x=umap_lab_pos[1,], y=umap_lab_pos[2,], label=colnames(umap_lab_pos), colour="grey35")
dev.off()
png(paste(name, out_tag, "_default_umap.png", sep="_"), width=6, height=6, units="in", res=100)
Seurat::DimPlot(myseur, reduction = "umap")
dev.off()
png(paste(name, out_tag, "perMT.png", sep="_"), width=6, height=6, units="in", res=150)
Seurat::FeaturePlot(myseur, "percent.mt", reduction="umap")
dev.off()
png(paste(name, out_tag, "nFeature.png", sep="_"), width=6, height=6, units="in", res=150)
Seurat::FeaturePlot(myseur, "nFeature_RNA", reduction="umap")
dev.off()
png(paste(name, out_tag, "CCphase.png", sep="_"), width=6, height=6, units="in", res=150)
Seurat::DimPlot(myseur, group.by="Phase", reduction="umap")
dev.off()
png(paste(name, out_tag, "sample.png", sep="_"), width=6, height=6, units="in", res=150)
Seurat::DimPlot(myseur, group.by="donor", reduction="umap")
dev.off()
SoupX_outfile <- paste(paste("SoupX/", name, sep=""), out_tag, "SoupX.rds", sep="_")
if (!file.exists(SoupX_outfile)) {
# SoupX
## Need to make the EmptyDrops version of this!!!
require("SoupX")
require("Seurat")
set.seed(this_seed)
# Create SoupX object
raw <- Read10X(paste(folder, "raw_gene_bc_matrices/GRCh38", sep="/"))
raw <- raw[,Matrix::colSums(raw) > 10]
raw <- raw[rownames(raw) %in% rownames(myseur),]
keep_cells <- colnames(myseur);
tot_umi <- Matrix::colSums(raw);
tot_is_cell <- min(tot_umi[colnames(raw) %in% colnames(myseur)])
raw_is_cell <- raw[,keep_cells]
myseur <- myseur[,match(colnames(raw_is_cell), colnames(myseur))]
sc <- SoupChannel(raw, raw_is_cell, [email protected], soupRange=c(10, tot_is_cell*0.9))
anno_cluster <- cell_anno_to_cluster_anno(as.character(unlist([email protected]$consistent_labs)), [email protected]$seurat_clusters)
anno_cluster <- anno_cluster$lab[match([email protected]$seurat_clusters, anno_cluster$cluster)]
sc <- setClusters(sc, anno_cluster);
sc <- setDR(sc, cbind(myseur@[email protected][,1], myseur@[email protected][,2]));
# Use known genes to estimate contamination fraction per cell
non_expressed_gene_list = list(HB =c("HBB", "HBA1", "HBA"),
TCR=c("TRAC", "TRBC1", "TRBC2", "TRDC", "TRGC1", "TRGC2"),
Drug=c("CYP2E1", "CYP3A7", "CYP2A7", "CYP2A6"),
BCR=c("IGKC", "IGHE", "IGHM", "IGLC1", "IGLC3", "IGLC2"),
Clot=c("F10", "F5", "F9", "F7", "FGB"),
Bile=c("SLC10A1", "SLC01B1", "SLCO1B3", "SLC22A1", "SLC22A7",
"ABCB1", "ABCB11", "ABCB4", "ABCC3", "ABCC6"));
non_expressed_gene_list <- lapply(non_expressed_gene_list, function(x){return(x[x %in% rownames(myseur)])})
useToEst = estimateNonExpressingCells(sc, nonExpressedGeneList = non_expressed_gene_list, clusters=NULL)
sc = calculateContaminationFraction(sc, non_expressed_gene_list, useToEst = useToEst, cellSpecificEstimates=TRUE)
quantile(sc$metaData$rho)
# Correcting the expression matrix.
out = adjustCounts(sc)
out <- out[,match(colnames(myseur), colnames(out))]
myseur@assays$SoupCorrected <- out;
saveRDS(myseur, SoupX_outfile);
} else {
myseur <- readRDS(SoupX_outfile)
}
#myseur <- do_annotation(myseur)
#orig_labels <- [email protected]
labels <- list()
agreement <- c()
myseur@assays$Orig_Counts <- myseur@assays$RNA@counts
# Compare autoannotation across corrections.
if (length(myseur@assays) >2) {
for (i in 2:length(myseur@assays)) {
this_corr_mat <- myseur@assays[[i]]
myseur<- myseur[rownames(myseur) %in% rownames(this_corr_mat),]
myseur<- myseur[, colnames(myseur) %in% colnames(this_corr_mat)]
this_corr_mat <- this_corr_mat[match(rownames(myseur), rownames(this_corr_mat)),]
this_corr_mat <- this_corr_mat[,match(colnames(myseur), colnames(this_corr_mat))]
myseur@assays$RNA@counts <- this_corr_mat
myseur <- Seurat::NormalizeData(myseur);
myseur <- Seurat::ScaleData(myseur);
myseur <- do_annotation(myseur);
corr_name <- names(myseur@assays)[i];
labels[[corr_name]] <- [email protected]
# plot
png(paste(name, out_tag, corr_name, "expression.png", sep="_"), width=6, height=6, units="in", res=100)
Seurat::FeaturePlot(myseur, features=c("ALB","APOB", "SLC22A1","HBB", "HBB", "APOA1", "MARCO", "TNF"))
dev.off()
myseur[["percent.mt"]] <- Seurat::PercentageFeatureSet(myseur, pattern = "^MT-")
myseur <- subset(myseur, subset = nFeature_RNA > ng_filter & percent.mt < mt_filter)
png(paste(name, out_tag, corr_name, "MT.png", sep="_"), width=6, height=6, units="in", res=100)
Seurat::FeaturePlot(myseur, features=c("percent.mt"))
dev.off()
#agreement of autoannotation labels
is.lab <- [email protected][,"general_labs"] != "ambiguous" &
[email protected][,"marker_general_labs"] != "ambiguous" &
[email protected][,"marker_general_labs"] != "Hepatocyte"
score <- sum([email protected][is.lab,"general_labs"] == [email protected][is.lab,"marker_general_labs"])/sum(is.lab)
agreement[corr_name] <- score;
}
}
png(paste(name, out_tag, corr_name, "Score.png", sep="_"), width=8, height=4, units="in", res=100)
par(mar=c(4,10,1,1))
barplot(agreement, horiz=TRUE, xlab="Autoannotation Agreement", col="grey65")
dev.off()
print(agreement)