#!/usr/bin/Rscript
require(optparse)
require(data.table)
require(rvatk)
# GLOBAL VARS
kOutputTypes <- c("qualvar","long","collapsing.matrix")
Main <- function() {
option.list <- GetOptList()
opt <- parse_args(OptionParser(option_list=option.list))
geno <- ReadLargeTable(opt$genotypes.csv,
showProgress=F,
header=T, check.names=F)
# read in sample file
sampleped <- read.table(opt$sampleped.file, stringsAsFactors=F)
cases <- sampleped[sampleped[,6]==2, 2]
ctrls <- sampleped[sampleped[,6]==1, 2]
samples <- c(cases, ctrls)
# verify no repeats of sampleID
if (TRUE %in% duplicated(samples)) {
stop("duplicated samples in input sample file")
}
# read in qualvar file if defined, else read in genotypes CSV
geno[[opt$gene.name.col]] <- gsub("'","", geno[[opt$gene.name.col]])
cnd <- ReadConditionFile(opt$condition.file)
if (nrow(cnd) > 0) {
if ("loo maf" %in% cnd[,1]) {
cat("Adding loo maf col...\n")
geno <- AddLooMafCol(geno, length(samples))
}
geno <- TblCndSubset(geno, cnd)
}
# make sure that only sample from sampleped are included in geno
geno <- geno[ geno[[opt$sample.name.col]] %in% samples, ]
allele.count <- GetAlleleCounts(geno[[opt$sample.name.col]],
geno[[opt$variant.id.col]],
geno[[opt$genotype.col]],
sampleped)
if (opt$output.type == "collapsing.matrix") {
if (is.null(opt$geneset.file)) {
cat("to create collapsing matrix output need geneset.\n")
q()
}
# read in geneset file
geneset <- ReadListFile(opt$geneset.file)
# verify no duplicated genes in geneset
if (TRUE %in% duplicated(geneset)) {
stop("duplicated gene names in input geneset file")
}
# build collapsing matrix based on geneset and sample files
mat <- CollapsingMatrixInit(geneset, samples)
qualvar <- cbind(geno[, c(opt$gene.name.col, opt$variant.id.col,
opt$sample.name.col), drop=F], allele.count)
mat <- CollapsingMatrixRecode(mat,
qualvar)
WriteLargeTable(mat, opt$output.file,
col.names=T,
row.names=F,
quote=F, sep="\t")
} else if (opt$output.type == "long") {
output <- geno[, c(opt$gene.name.col, opt$variant.id.col,
opt$sample.name.col), drop=F]
} else {
qualvar <- cbind(geno[, c(opt$gene.name.col, opt$variant.id.col,
opt$sample.name.col), drop=F], allele.count)
if (is.null(opt$geneset.file)==F) {
geneset <- ReadListFile(opt$geneset.file)
qualvar <- qualvar[qualvar[,1] %in% geneset, ]
}
WriteLargeTable(qualvar, opt$output.file,
col.names=F,
row.names=F, quote=F, sep="\t")
}
}
GetOptList <- function() {
# Get option list.
#
# Returns:
# Option list to parse options.
option.list <- list(
make_option(c("--genotypes-csv"), action="store", type="character",
default=NULL, dest="genotypes.csv",
help="name of genotypes csv generated by ATAV,
if no qualvar file provided, required, [default %default]"),
make_option(c("--output-file"),
action="store", type="character",
default=NULL,
dest="output.file",
help="output file name, required, [default %default]"),
make_option(c("--output-type"),
action="store", type="character",
default="collapsing.matrix",
dest="output.type",
help="format for output file, [default %default]"),
make_option(c("--sampleped-file"), action="store", type="character",
default=NULL, dest="sampleped.file",
help="name of required sample file, [default %default]"),
make_option(c("--geneset-file"), action="store",
type="character",
default=NULL,
dest="geneset.file",
help="name of required geneset file, [default %default]"),
make_option(c("--condition-file"), action="store", type="character",
default=NULL, dest="condition.file",
help="name of condition file to subset genotype csv on,
if no qualvar file is provided [default %default]"),
make_option(c("--gene-name-col"), action="store", type="character",
default="Gene Name", dest="gene.name.col",
help="Gene name column to subset genotype csv on,
if no qualvar file is provided [default %default]"),
make_option(c("--variant-id-col"), action="store", type="character",
default="Variant ID", dest="variant.id.col",
help="variant ID column to subset genotype csv on,
if no qualvar file is provided [default %default]"),
make_option(c("--sample-name-col"), action="store", type="character",
default="Sample Name", dest="sample.name.col",
help="Sample name column to subset genotype csv on,
if no qualvar file is provided [default %default]"),
make_option(c("--genotype-col"), action="store", type="character",
default="Genotype", dest="genotype.col",
help="Sample name column to subset genotype csv on,
if no qualvar file is provided [default %default]")
)
return(option.list)
}
if (interactive() == F) {
Main()
}
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.