chipster.tools.path = '/opt/chipster/tools'
chipster.common.path = '/opt/chipster/comp/modules/common/R-2.12'
chipster.module.path = '/opt/chipster/comp/modules/microarray'
setwd("77a3fe54-40cc-46a1-99c9-7f390c2193be")
image.width <- 600
image.height <- 600
# TOOL ordination-nmds.R: NMDS (Non-metric multidimensional scaling. Creates a 2-dimensional representation of the arrays using Euclidean distances. Can be used for quality control.)
# INPUT normalized.tsv: normalized.tsv TYPE GENE_EXPRS 
# INPUT META phenodata.tsv: phenodata.tsv TYPE GENERIC 
# OUTPUT nmds.pdf: nmds.pdf 
# PARAMETER image.width: image.width TYPE INTEGER FROM 200 TO 3200 DEFAULT 600 (Width of the plotted network image)
# PARAMETER image.height: image.height TYPE INTEGER FROM 200 TO 3200 DEFAULT 600 (Height of the plotted network image)


# Non-metric multidimensional scaling
# JTT 14.1.2008

# MG, 15.9.2011
# Updated colors and legend

# Parameter settings (default) for testing purposes
#image.width<-c(600)
#image.height<-c(600)

# Renaming variables
w<-image.width
h<-image.height

# Loads the libraries
library(MASS)

# Loads the normalized data
file<-c("normalized.tsv")
dat<-read.table(file, header=T, sep="\t", row.names=1)

# Separates expression values and flags
calls<-dat[,grep("flag", names(dat))]
dat2<-dat[,grep("chip", names(dat))]

# Loads phenodata
phenodata<-read.table("phenodata.tsv", header=T, sep="\t")

# Calculating the distance matrix
dat2.dist<-dist(t(dat2))

# Calculating the MDS
mds<-isoMDS(dat2.dist)


# Setup sample colors according to group
if (length(levels(as.factor(phenodata$group))) > 0) {
	sample_colors <- numeric(length(phenodata$group))
	group_levels <- levels(as.factor(phenodata$group))
	group_identity <- as.character(phenodata$group)
	for (count_levels in 1:length(group_levels)) {
		for (count_samples in 1:length(phenodata$group)) {
			if(group_identity[count_samples]==group_levels[count_levels]) sample_colors[count_samples] <- 1+count_levels
		}
	}
	level_colors <- levels(as.factor(sample_colors))
}
if (length(levels(as.factor(phenodata$group))) == 0) {
	sample_colors <- rep(2,length(phenodata$group))
}


# Plotting the image
pdf(file="nmds.pdf", width=w/72, height=h/72)
plot(mds$points[,1], mds$points[,2], main="NMDS", pch=19, xlab="Dimension 1", ylab="Dimension 2", type="n")
text(mds$points[,1], mds$points[,2], phenodata$description, cex=0.75, col=sample_colors)
if (length(levels(as.factor(phenodata$group))) > 0) {
	legend (x="topleft", legend=group_levels, col=level_colors, cex=0.5, pch=19)
}
dev.off()
print("script-finished-succesfully")
