extending dpclust master function to read from bb rho_psi files and a…

…dding an example pipeline

R/preprocessing.R

@@ -738,19 +738,29 @@ addVcfInfoCol = function(vcf, data, number, type, description, abbreviation) {
 #' @param indeldatafiles A vector with indel DPClust input files (Default: NULL)
 #' @author sd11
 #' @export
-createProjectFile = function(outputfile, donornames, samplenames, purities, sex, datafiles=paste(samplenames, "_allDirichletProcessInfo.txt", sep=""), cndatafiles=NULL, indeldatafiles=NULL) {
-  if (!all(c(length(samplenames), length(purities), length(sex), length(datafiles))==length(donornames))) {
-    stop("All provided vectors must be of the same length")
+createProjectFile = function(outputfile, donornames, samplenames, sex, purities=NULL, rho_and_psi_files=NULL, datafiles=paste(samplenames, "_allDirichletProcessInfo.txt", sep=""), cndatafiles=NULL, indeldatafiles=NULL) {
+  if (is.null(purities) & is.null(rho_and_psi_files)) {
+    stop("Please provide either a vector of purities or a vector of Battenberg rho_and_psi files")
   }
-  if (!is.null(cndatafiles)) {
-    if (length(cndatafiles)!=length(donornames)) {
-      stop("All provided vectors must be of the same length")
+
+  .checklength = function(a) {
+    if (!is.null(a)) {
+      if (length(a)!=length(donornames)) {
+        stop("All provided vectors must be of the same length")
+      }
     }
   }
-  if (!is.null(indeldatafiles)) {
-    if (length(indeldatafiles)!=length(donornames)) {
-      stop("All provided vectors must be of the same length")
-    }
+  .checklength(donornames)
+  .checklength(samplenames)
+  .checklength(sex)
+  .checklength(purities)
+  .checklength(rho_and_psi_files)
+  .checklength(datafiles)
+  .checklength(indeldatafiles)
+  .checklength(cndatafiles)
+
+  if (!is.null(rho_and_psi_files)) {
+    purities = unlist(lapply(rho_and_psi_files, GetCellularity))
   }
 
   output = data.frame(sample=donornames, 

inst/example/preproc_dpclust_master_file.R

@@ -0,0 +1,46 @@
+#'
+#' Simple pipeline that produces a DPClust master file
+#' 
+#' v1.0 - 2018-11-02 - sd11 [at] sanger.ac.uk
+
+library(dpclust3p)
+library(optparse)
+
+option_list = list(
+  make_option(c("-s", "--samplenames"), type="character", default=NULL, help="Comma separated list of samplenames", metavar="character"),
+  make_option(c("-d", "--donornames"), type="character", default=NULL, help="Comma separated list of donornames (use the same donor name to match multiple samples)", metavar="character"),
+  make_option(c("-p", "--purities"), type="character", default=NULL, help="Comma separated list of purities (supply -p or -r)", metavar="character"),
+  make_option(c("-r", "--rho_and_psi"), type="character", default=NULL, help="List of Battenberg rho and psi output files (supply -p or -r)", metavar="character"),
+  make_option(c("--sex"), type="character", default=NULL, help="Comma separated list of sex of the sample", metavar="character"),
+  make_option(c("-o", "--output"), type="character", default=NULL, help="Output file", metavar="character")
+)
+
+opt_parser = OptionParser(option_list=option_list)
+opt = parse_args(opt_parser)
+
+donornames = strsplit(opt$donornames, ",")
+samplenames = strsplit(opt$samplenames, ",")
+sex = strsplit(opt$sex, ",")
+outputfile = opt$output
+
+if (!is.null(opt$rho_and_psi)) {
+  rho_and_psi_files = strsplit(opt$rho_and_psi, ",")
+} else {
+  rho_and_psi_files = NULL 
+}
+if (!is.null(opt$purities)) {
+  purities = strsplit(opt$purities, ",")
+} else {
+  purities = NULL
+}
+
+if (!all(sex %in% c("male", "female"))) {
+  stop("Provide male or female as sex")
+}
+
+createProjectFile(outputfile=outputfile, 
+                  donornames=donornames, 
+                  samplenames=samplenames, 
+                  sex=sex, 
+                  purities=purities, 
+                  rho_and_psi_files=rho_and_psi_files)