-
Notifications
You must be signed in to change notification settings - Fork 3
/
Figure1A_heatmaps.R
169 lines (142 loc) · 5.67 KB
/
Figure1A_heatmaps.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
source("~/MAGIC/R_Imputation_Functions.R")
Sim <- function(n_genes=500, n_cells=500, l10fc=1, disp=0.2, min_mean=-3, max_mean=6, propDE=0.5, type=c("clust", "time"), minor_type_freq=0.5, d_rate=0) {
mus <- runif(n_genes, min=min_mean+0.01, max=max_mean-0.01)
up <- 1:(n_genes*propDE/2)
down <- (max(up)+1):((n_genes*propDE))
mus2 <- mus
mus2[up] <- mus[up]+l10fc
mus2[down] <- mus[down]-l10fc
mus2[mus2 > max_mean] <- max_mean
mus2[mus2 < min_mean] <- min_mean
time <- runif(n_cells);
n_cells1 <- n_cells * (1-minor_type_freq)
n_cells2 <- n_cells * minor_type_freq
mu_tab <- 10^mus %*% t(rep(1, times=n_cells))
if (type[1]=="clust") {
mu_tab2 <- 10^mus2 %*% t(rep(1, times=n_cells2))
truth <- rep(0, time=n_cells);
mu_tab[,1:n_cells2] <- mu_tab2;
truth[1:n_cells2] <- 1
} else if (type[1]=="time") {
mu_tab2 <- 10^mus2 %*% t(rep(1, times=n_cells))
mu_tab <- t( t(mu_tab2)*(time) + t(mu_tab)*(1-time) )
truth=time
}
MAT <- sapply(1:n_genes, function(i) {
sapply(mu_tab[i,], function(m) {
if(runif(1) > d_rate) {rnbinom(1, mu=m, size=1/disp)}
else{0};
})
})
MAT <- t(MAT)
is.up <- 1:n_genes %in% up
is.down <- 1:n_genes %in% down
keep_g <- rowSums(MAT) > 0
MAT <- MAT[keep_g,]
is.up <- is.up[keep_g]
is.down <- is.down[keep_g]
keep_c <- colSums(MAT) > 0
MAT <- MAT[,keep_c]
truth <- truth[keep_c]
rownames(MAT) <- paste("g", 1:nrow(MAT), sep="");
colnames(MAT) <- paste("c", 1:ncol(MAT), sep="");
return(list(mat=MAT, g_up=is.up, g_down=is.down, cell_truth=truth))
}
gene_cor_heatmap <- function(sim, mat="mat") {
require("scater")
this_mat <- as.matrix(sim@assays[[mat]])
cor_mat <- cor(t(this_mat), method="spearman")
require("gplots")
require("RColorBrewer")
heat_cols <- rev(brewer.pal(8, "RdBu"))
bin_edges <- seq(from=-1, to=1, length=9)
gene_info <- data.frame(is.up=rowData(sim)$g_up, is.down=rowData(sim)$g_down, expr=rowMeans(this_mat))
reorder <- order(gene_info[,1], gene_info[,2], gene_info[,3], decreasing=T)
gene_info <- gene_info[reorder,]
sideCols <- rep("grey85", nrow(sim))
sideCols[gene_info[,1]] <- "red"
sideCols[gene_info[,2]] <- "blue"
cor_mat <- cor_mat[reorder, reorder]
heatmap.2(cor_mat, col=heat_cols, trace="none", Rowv=FALSE, Colv=FALSE, dendrogram="none", ColSideColors=sideCols, RowSideColors=sideCols, breaks=bin_edges)
invisible(cor_mat)
}
### Generate Simulated Data ###
require("scater")
set.seed(20194)
sims <- Sim(n_genes=5000, n_cells=1000, max_mean=4, min_mean=-3, propDE=0.5)
sims$counts <- sims$mat;
sf <- colSums(sims$mat)
sims$norm <- t(t(sims$mat)/sf*median(sf))
sims$lognorm <- log2(sims$norm+1)
cdat <- data.frame(cell_truth=sims$cell_truth)
rdat <- data.frame(g_up=sims$g_up, g_down=sims$g_down)
rownames(rdat) <- rownames(sims$counts)
rownames(cdat) <- colnames(sims$counts)
cdat$Group <- cdat$cell_truth
sim_sce <- SingleCellExperiment(assays=list(counts=sims$counts, logcounts=sims$lognorm), colData=cdat, rowData=rdat)
saveRDS(sim_sce, "Heatmap_sim_object.rds")
### Impute it. ###
require("Rmagic")
require("DrImpute")
require("scImpute")
require("SAVER")
source("/nfs/users/nfs_t/ta6/MAGIC/knn_smooth.R")
set.seed(28198)
n.cores=16
res <- scImpute_wrapper(sim_sce, n.cores=n.cores, do.norm=FALSE)
assays(sim_sce)[["sci"]] <- res;
res <- DrImpute_wrapper(sim_sce, do.norm=FALSE)
assays(sim_sce)[["dri"]] <- res;
res <- MAGIC_wrapper(sim_sce, do.norm=FALSE)
assays(sim_sce)[["magic"]] <- res;
res <- knn_wrapper(sim_sce, do.norm=FALSE)
assays(sim_sce)[["knn"]] <- res;
res <- SAVER_wrapper(sim_sce, n.cores=n.cores, do.norm=FALSE)
assays(sim_sce)[["saver"]] <- res;
saveRDS(sim_sce, file="Heatmap_sim_object.rds") # Add autoencoders to this RDS separately
### Heatmap ###
sims_sce <- readRDS("Heatmap_sim_object.rds") # Add autoencoders to this RDS separately
set.seed(3819)
subset <- sample(1:nrow(sims_sce), min(500, nrow(sims_sce))) #Subset genes to make visualization more legible
sims_sce <- sims_sce[subset,]
png(paste("Raw_GeneCor_heatmap3.png", sep=""), width=4, height=4, units="in", res=300);
gene_cor_heatmap(sims_sce, "counts")
dev.off()
png(paste("RawLog_GeneCor_heatmap3.png", sep=""), width=4, height=4, units="in", res=300);
gene_cor_heatmap(sims_sce, "logcounts")
dev.off()
png(paste("MAGIC_GeneCor_heatmap3.png", sep=""), width=4, height=4, units="in", res=300);
gene_cor_heatmap(sims_sce, "magic")
dev.off()
png(paste("DrImpute_GeneCor_heatmap3.png", sep=""), width=4, height=4, units="in", res=300);
gene_cor_heatmap(sims_sce, "dri")
dev.off()
png(paste("scImpute_GeneCor_heatmap3.png", sep=""), width=4, height=4, units="in", res=300);
gene_cor_heatmap(sims_sce, "sci")
dev.off()
png(paste("SAVER_GeneCor_heatmap3.png", sep=""), width=4, height=4, units="in", res=300);
gene_cor_heatmap(sims_sce, "saver")
dev.off()
png(paste("DCA_GeneCor_heatmap3.png", sep=""), width=4, height=4, units="in", res=300);
gene_cor_heatmap(sims_sce, "dca")
dev.off()
png(paste("scVI_GeneCor_heatmap3.png", sep=""), width=4, height=4, units="in", res=300);
gene_cor_heatmap(sims_sce, "scVI")
dev.off()
png(paste("Knn_GeneCor_heatmap3.png", sep=""), width=4, height=4, units="in", res=300);
gene_cor_heatmap(sims_sce, "knn")
dev.off()
source("Colour_bar.R")
blank_plot <- function() {
tmp <- par("mar")
par(mar=c(0,0,0,0))
plot(1,1, col="white", xlim=c(0,1), ylim=c(0,1), xaxt="n", yaxt="n", main="", xlab="", ylab="", bty="n")
return(tmp);
}
heat_cols <- rev(brewer.pal(8, "RdBu"))
bin_edges <- seq(from=-1, to=1, length=9)
png("GeneCor_heatmap3_colorbar.png", width=7, height=3, units="in", res=300)
blank_plot()
color.bar(heat_cols, min=-1, max=1, ticks.at=bin_edges, title="Correlation", horiz=T, add=T)
dev.off()
###