annotate cluster.tools/remove.tcga.normals.R @ 9:a3c03541fe6f draft default tip

Uploaded
author peter-waltman
date Mon, 11 Mar 2013 17:30:48 -0400
parents 563832f48c08
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
1 #!/usr/bin/env Rscript
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
2 argspec <- c("remove.tcga.normals.R removes TCGA normal samples from a given matrix
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
3
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
4 Usage:
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
5 remove.tcga.normals.R -d <data.file>
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
6 Optional:
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
7 -o <output_file>
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
8 \n\n")
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
9 args <- commandArgs(TRUE)
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
10 if ( length( args ) == 1 && args =="--help") {
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
11 write(argspec, stderr())
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
12 q();
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
13 }
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
14
3
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
15 ## some helper fn's
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
16 write.2.tab <- function( mat,
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
17 fname ) {
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
18 mat <- rbind( colnames( mat ), mat )
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
19 mat <- cbind( c( "ID", rownames( mat )[-1] ),
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
20 mat )
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
21 write.table( mat, fname, sep="\t", row.names=FALSE, col.names=FALSE, quote=FALSE )
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
22 }
0
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
23 lib.load.quiet <- function( package ) {
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
24 package <- as.character(substitute(package))
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
25 suppressPackageStartupMessages( do.call( "library", list( package=package ) ) )
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
26 }
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
27 lib.load.quiet(getopt)
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
28
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
29 spec <- matrix( c( "data.fname", "d", 1, "character",
3
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
30 "output.fname", "o", 2, "character",
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
31 "return.normals", "r", 2, "character",
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
32 "out.norm.fname", "O", 2, "character"
0
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
33 ),
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
34 nc=4,
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
35 byrow=TRUE
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
36 )
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
37
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
38 opt <- getopt( spec=spec )
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
39 if ( is.null( opt$output.fname ) ) { opt$output.fname <- 'merge_merge.tumors.tab' }
3
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
40 if ( is.null( opt$return.normals ) ) {
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
41 opt$return.normals <- FALSE
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
42 } else {
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
43 opt$return.normals <- ( tolower( opt$return.normals ) %in% "yes" )
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
44 }
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
45 if ( is.null( opt$out.norm.fname ) ) { opt$out.norm.fname <- 'merge_merge.normals.tab' }
0
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
46
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
47 mat <- as.matrix( read.delim( opt$data.fname, row.names=1, check.names=FALSE ) )
3
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
48 norms <- matrix( NA, nc=0, nr=nrow( mat ), dimnames=list( rownames( mat ), c() ) )
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
49 if ( length( strsplit( colnames( mat ), "-" )[[1]] ) > 3 ) {
0
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
50 cnames <- sapply( strsplit( colnames( mat ), "-" ), function(x) x[4] )
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
51 norms <- grepl( "^1", cnames )
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
52
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
53 if ( sum( norms ) > 0 ) {
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
54 tumors <- ! norms
3
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
55 norms <- mat[, norms ]
0
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
56 mat <- mat[, tumors ]
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
57 }
3
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
58 } else {
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
59 if ( opt$return.normals ) {
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
60 writeLines( "TCGA ID barcodes in supplied file only provide patient sample info (no aliquot components are in IDs)\n", opt$out.norm.fname )
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
61 }
0
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
62 }
3
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
63 write.2.tab( mat, opt$output.fname )
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
64 if ( opt$return.normals ) {
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
65 if ( ncol( norms ) > 0 ) {
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
66 write.2.tab( norms, opt$out.norm.fname )
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
67 } else {
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
68 writeLines( "no normals found in supplied matrix\n", opt$out.norm.fname )
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
69 }
563832f48c08 Uploaded
peter-waltman
parents: 0
diff changeset
70 }