annotate cluster.tools/impute.knn.R @ 0:0decf3fd54bc draft

Uploaded
author peter-waltman
date Thu, 28 Feb 2013 01:45:39 -0500
parents
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("impute.knn.R replaces missing values, using the impute.knn function from the impute package
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 impute.knn.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
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
15 lib.load.quiet <- function( package ) {
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
16 package <- as.character(substitute(package))
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
17 suppressPackageStartupMessages( do.call( "library", list( package=package ) ) )
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
18 }
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
19 lib.load.quiet(getopt)
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
20 lib.load.quiet( impute )
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
21
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
22 spec <- matrix( c( "data.fname", "d", 1, "character",
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
23 "output.fname", "o", 2, "character"
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
24 ),
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
25 nc=4,
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
26 byrow=TRUE
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
27 )
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
28
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
29 opt <- getopt( spec=spec )
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
30
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
31 data <- as.matrix( read.delim( opt$data.fname, row.names=1, check.names=FALSE ) )
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
32 if ( is.null( opt$replacement.val ) ) { opt$replacement.val <- NA }
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
33 if ( is.null( opt$output.fname ) ) { opt$output.fname <- paste( "impute.knn", basename( opt$data.fname ), sep="." ) }
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
34
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
35 ## Set any NA, NAN or Inf entries to 0
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
36 if ( is.nan(data) ) {
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
37 data[ is.nan( data ) ] <- NA
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
38 }
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
39 if ( is.infinite(data) ) {
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
40 data[ is.infinite( data ) ] <- NA
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
41 }
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
42
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
43 data <- impute.knn( data )$data
0decf3fd54bc Uploaded
peter-waltman
parents:
diff changeset
44 write.table( data, opt$output.fname, sep="\t", quote=FALSE, col.names=NA )