Merge branch 'master' into remove_dup_NULLs

analyses/immune-deconv/02-summary-plots.R

@@ -14,16 +14,14 @@ source(file.path(root_dir, "analyses", "immune-deconv",
                  "util", "pubTheme.R"))
 
 option_list <- list(
-  make_option(c("-i", "--input"), type = "character",
-              help = "Immunedeconv output from 01-immune.deconv.R (.RData)"),
-  make_option(c("-o", "--output"), type = "character",
-              help = "Output PDF (.pdf)")
+  make_option(c("-i", "--input"), type = "character", help = "Immunedeconv output from 01-immune.deconv.R (.RData)"), 
+  make_option(c("-o",  "--output"), type = "character", help = "Output directory")
 )
 
 # Example Run:
 # Rscript analyses/immune-deconv/02-summary-plots.R \
 # -i 'analyses/immune-deconv/results/deconv-output.RData' \
-# -o 'analyses/immune-deconv/results/deconv-summary.pdf'
+# -o 'analyses/immune-deconv/plots
 
 # parse parameters
 opt <- parse_args(OptionParser(option_list = option_list))
@@ -111,18 +109,29 @@ total <- total %>%
   column_to_rownames('cell_type') %>%
   replace(is.na(.), 0)
 
-# create correlation heatmap
-pdf(file = output, onefile = TRUE, width = 13, height = 8)
+# replace space from method names for output filename
+m1 <- gsub(" ","",method1.name) 
+m2 <- gsub(" ","",method2.name)
+
+# create correlation plot for overlapping cell types between both methods
+png(filename = file.path(output, paste0("corrplot_", m1, "_vs_", m2, ".png")), 
+    width = 13, height = 8, units = "in", res = 300)
 corrplot(t(total), method = "circle", type = 'full', win.asp = 0.5, 
          addCoef.col = "black", number.cex = .5,
          is.corr = FALSE, tl.cex = 0.8, mar = c(0, 0, 0, 5), 
          title = paste0("\n\n\n\nCorrelation matrix (", 
                         method1.name, " vs ", method2.name, ")\n",
                         "Overall Pearson Correlation: ", avg.cor))
+dev.off()
 
-# lastly, create heatmaps for both deconvolution methods
-# add to the same file as above
+# create heatmaps of average immune scores per histology per cell type
+# method1
+png(filename = file.path(output, paste0("heatmap_", m1, ".png")), width = 13, height = 8, units = "in", res = 300)
 create.heatmap(deconv.method = method1, title = method1.name)
+dev.off()
+
+# method2
+png(filename = file.path(output, paste0("heatmap_", m2, ".png")), width = 10, height = 8, units = "in", res = 300)
 create.heatmap(deconv.method = method2, title = method2.name)
 dev.off()
 

analyses/immune-deconv/README.md

@@ -62,11 +62,9 @@ This script creates summary correlation plots and heatmaps from predicted immune
 
 3. Output
 
-`plots/deconv-summary.pdf`
-
-The pdf contains the following plots:
-* a correlation plot of predicted immune scores across 13 common cell types from xCell and CIBERSORT (abs.) 
-* individual heatmaps of average immune scores per cell type per histology for each method. 
+* `plots/corrplot_xCell_vs_CIBERSORT(abs.).png`: a correlation plot of predicted immune scores across 13 common cell types from xCell and CIBERSORT (abs.) 
+* `plots/heatmap_xCell.png`: heatmap of average immune scores per cell type per histology for xCell.
+* `plots/heatmap_CIBERSORT(abs.).png`: heatmap of average immune scores per cell type per histology for the second method, CIBERSORT in our case.
 
 ### Running the analysis
 

analyses/immune-deconv/run-immune-deconv.sh

@@ -52,4 +52,5 @@ echo "Create summary plots"
 # Also generates corresponding heatmaps of average normalized immune scores
 Rscript --vanilla 02-summary-plots.R \
 --input results/deconv-output.RData \
---output plots/deconv-summary.pdf
+--output plots
+