expdat <- read.table(file="gene2go.mm.6Dec06.txt",header=FALSE,sep="\t",quote="")
expgene <- unique(expdat$V2)
totgene <- length(expgene)
expterm <- unique(expdat$V3)

#calculate univariate frequencies
count <- numeric()
for (i in 1:length(expterm))
	{obsb <- expdat[expdat$V3==expterm[i],]
	obsb <- unique(obsb$V2)
	count <- append(count,length(obsb))}
expcounts <- data.frame(expterm,count)
write.table(expcounts,"expcounts.txt",row.names=FALSE,sep="\t",col.names=FALSE)

#create matrix of joint counts
vec <- vector(mode="numeric",length=length(expterm))
for (j in 1:length(expterm))
	{expc <- expdat[expdat$V3==expterm[1] | expdat$V3==expterm[j],2:3]
	expd <- unique(expc)
	expe <- unique(expd$V2)
	vec[j] <- dim(expd)[1] - length(expe)}
dat <- vec
dat <- t(dat)
write(dat,file="joint.txt",ncolumns=length(expterm))
for (i in 2:length(expterm))
	{vec <- vector(mode="numeric",length=length(expterm))
	for (j in 1:length(expterm))
		{expc <- expdat[expdat$V3==expterm[i] | expdat$V3==expterm[j],2:3]
		expd <- unique(expc)
		expe <- unique(expd$V2)
		vec[j] <- dim(expd)[1] - length(expe)}
	dat <- read.table(file="joint.txt")
	dat <- rbind(dat,vec)
	dat <- t(dat)
	write(dat,file="joint.txt",ncolumns=length(expterm))}

#calculate covariances
#expcounts <- read.table(file="expcounts.txt",header=FALSE)
#for (i in 1:length(expterm))
#	{for (j in 1:length(expterm))
#		{dat2[i,j] <- #dat2[i,j]/totgene-((expcounts[expcounts$V1==expterm[i],2]/totgene)*(expcounts[expcounts$V1==expterm[j],2]/totgene))}}

#calculate variances
#vari <- vector(mode="numeric",length=length(expterm))
#for (i in 1:length(expterm))
#	{vari[i] <- #(expcounts[expcounts$V1==expterm[i],2]/totgene)*((totgene-expcounts[expcounts$V1==expterm[i],2])/totgene)}

#calculate correlations
#for (i in 1:length(expterm))
#	{for (j in 1:length(expterm))
#		{if (i==j) {dat2[i,j] <- 1.0} else {dat2[i,j] <- dat2[i,j]/sqrt(vari[i]*vari[j])}}}
#dat2 <- t(dat2)
#write(dat2,file="corr.txt")

#see correl.c for converting the joint probability matrix into the correlation matrix