# Plot oncoprint of CoMMpass data CGA results (Figure 6G)
library(ComplexHeatmap)
library(circlize)
library(RColorBrewer)
library(grid)
library(dplyr)
library(viridis)
# load data
load("MM_COMPASS_FM.Rdata")
load("GSVA_MM_COMPASS_scores.Rdata")
# read correlation results
cor <- read.table("TableS6_CoMMpass_MM_nCGA_correlations.tsv", sep = "\t", header = TRUE)
# only samples with gexp data
fm <- fm[, !is.na(fm["N:GEXP:KRAS",])]
# add subtypes
subtypes <- rep("CCND1", length(fm))
subtypes[fm["B:SAMP:cancermap_subtypes_WHSC1_FGFR3_Ig",]==1] <- "FGFR3"
subtypes[fm["B:SAMP:cancermap_subtypes_Hyperdiploid",]==1] <- "Hyperdiploid(gain(11q)"
subtypes[fm["B:SAMP:cancermap_subtypes_Hyperdiploid_amp1q",]==1] <- "Hyperdiploid/gain(1q)"
subtypes[fm["B:SAMP:cancermap_subtypes_MAF_Ig",]==1] <- "MAF"
subtypes[fm["B:SAMP:cancermap_subtypes_TRAF3_Aberrated",]==1] <- "TRAF3"
subtypes[fm["B:SAMP:cancermap_cluster_CGA_Prolif",]==1] <- "CGA/Proliferative"
# order fm and viz by CGA number
cga_order <- order(fm["N:SAMP:nCGA",])
fm <- fm[,cga_order]
viz_scores <- viz_scores[,cga_order]
# create matrix with selected GSVA scores
feats <- c("MYC_TARGETS_V1-MSIGDB_HALLMARKS",
"DNA_REPLICATION-KEGG_MSIGDB_C2",
"E2F_TARGETS-MSIGDB_HALLMARKS",
"G2M_CHECKPOINT-MSIGDB_HALLMARKS",
"MITOTIC_M_M_G1_PHASES-REACTOME_MSIGDB_C2",
"INFLAMMATORY_RESPONSE-MSIGDB_HALLMARKS",
"TNFA_SIGNALING_VIA_NFKB-MSIGDB_HALLMARKS",
"BCR_SIGNALING_PATHWAY-SIG_MSIGDB_C2")
mat <- viz_scores[feats,]
mat <- mat[!is.na(mat[,1]),] # remove CGA genes not found
mat_scaled <- t(apply(mat, 1, scale))
colnames(mat_scaled) <- colnames(mat)
rownames(mat_scaled) <- gsub("_", " ",gsub("\\-.*", "", rownames(mat_scaled)))
# create matrix with selected genetic alteration features
alt <- data.frame(NRAS = as.character(fm["B:GNAB:NRAS",]),
`CCND1-Ig` = as.character(fm["B:CNVR:SeqWGS_CCND1_Ig_translocation",]),
`MAF-Ig` = as.character(fm["B:CNVR:SeqWGS_MAF_Ig_translocation",]))
colnames(alt) <- c("NRAS", "CCND1-Ig", "MAF-Ig")
# create heatmap annotations
annot <- data.frame(subtype = subtypes[cga_order])
gexp <- data.frame(`HLA II score` = as.numeric(fm["N:SAMP:HLAIIScore",]))
colnames(gexp) <- "HLA II score"
gexp_scaled <- as.data.frame(apply(gexp, 2, scale))
mat_hla2 <- t(gexp_scaled)
colnames(mat_hla2) <- colnames(mat)
ramp <- colorRamp2(seq(-2, 2, length.out = 11), rev(brewer.pal(11, "RdBu")))
green <- structure(c("darkgreen", "#EDEDED", "grey50"), names = c("1", "0", "NA"))
blackgrey <- structure(c("black", "#EDEDED", "grey50"), names = c("1", "0", "NA"))
mut <- as.numeric(fm["N:CLIN:NS_Non_IG_Variants",])
mut[mut > 150] <- 150 # mutation load cutoff due to hypermutated samples
ha1 <- HeatmapAnnotation(`# CGA` = anno_barplot(as.numeric(fm["N:SAMP:nCGA",]),
bar_width = 1,
border = FALSE,
axis = TRUE,
axis_param = list(gp = gpar(fontsize = 5, lwd = 0.5)),
gp = gpar(col = NA, fill = "grey30"),
ylim = c(0, 13)),
annotation_name_gp = gpar(fontsize = 10),
height = unit(1, "cm"))
ha2 <- HeatmapAnnotation(df = alt, col = list(NRAS = blackgrey,
`CCND1-Ig` = green,
`MAF-Ig` = green),
annotation_name_gp = gpar(fontsize = 10),
show_legend = F,
height = unit(1, "cm"))
ha3 <- HeatmapAnnotation(`# Hyperdiploid chr` = anno_barplot(as.numeric(fm["N:CNVR:Hyperdiploid_Chr_Count",]),
bar_width = 0.75,
border = FALSE,
axis = TRUE,
axis_param = list(gp = gpar(fontsize = 5, lwd = 0.5)),
gp = gpar(col = NA, fill = "grey20")),
Proliferation = anno_barplot(as.numeric(fm["N:CLIN:Prolif_Index",]),
bar_width = 0.75,
border = FALSE,
axis = TRUE,
axis_param = list(gp = gpar(fontsize = 5, lwd = 0.5)),
gp = gpar(col = NA, fill = "grey20")),
Mutations = anno_barplot(mut,
bar_width = 0.75,
border = FALSE,
axis = TRUE,
axis_param = list(gp = gpar(fontsize = 5, lwd = 0.5)),
gp = gpar(col = NA, fill = "grey20"),
ylim = c(0, 150)),
annotation_name_gp = gpar(fontsize = 10),
show_legend = F,
height = unit(1, "cm")
)
# make heatmaps
ht <- Heatmap(mat_scaled,
name = "oncoprint",
col = colorRamp2(seq(-1, 1, length.out = 11), viridis(11, begin = 0.1, end = 0.9)),
row_names_side = "right",
row_names_gp = gpar(fontsize = 6),
show_column_names = FALSE,
show_column_dend = FALSE,
cluster_columns = FALSE,
cluster_rows = FALSE,
show_row_dend = FALSE,
row_title_gp = gpar(fontsize = 8),
show_heatmap_legend = T,
heatmap_legend_param = list(title = "GSVA\nZ-score",
title_gp = gpar(fontsize = 10),
labels_gp = gpar(fontsize = 10),
grid_height = unit(0.2, "cm"),
grid_width = unit(2, "mm"),
legend_direction = "horizontal",
title_position = "topcenter")
)
ht_hla <- Heatmap(mat_hla2,
name = "HLA II score",
col = ramp,
row_names_side = "right",
row_names_gp = gpar(fontsize = 8),
show_column_names = FALSE,
show_column_dend = FALSE,
cluster_columns = FALSE,
cluster_rows = FALSE,
show_row_dend = FALSE,
show_heatmap_legend = T,
heatmap_legend_param = list(title = "HLA II\nZ-score",
title_gp = gpar(fontsize = 10),
labels_gp = gpar(fontsize = 10),
grid_height = unit(0.2, "cm"),
grid_width = unit(2, "mm"),
legend_direction = "horizontal",
title_position = "topcenter")
)
# combine heatmaps and annotations
ht_list <- ha1 %v% ha2 %v% ha3 %v% ht_hla %v% ht
# print oncoprint
pdf("Figure6G_CoMMpass_CGA_oncoprint.pdf", height = 3.5, width = 5)
draw(ht_list, padding = unit(c(2, 5, 2, 2), "mm"), heatmap_legend_side = "bottom")
dev.off()
Datasets
Models
