##########################################################################
# This program reads in two text files and output a simplied version of 
# table3 in the paper.
# 
# The two input files:
# vircotable3data.txt
# virologictable3data.txt
#
# Output file format:
# Refer to table3
#
# Usage:
# ftest(filename1, filename2)    ## filename should involve directory
#
# In this program, Fisher's exact test was done to test 
# whether the proportions of isolates with a pattern of drug resistance
# mutations that had levels of fold-resistance exceeding the biological cut-off
# were significantly different between the two assays. 
##########################################################################

ftest <- function(file1, file2) {
	## Read in virco and virologic data
	vircodata = read.table(file1, sep="\t", header=T)
	virologicdata = read.table(file2, sep="\t", header=T)
	
	## Get simplized files
	uniqVirco = unique(vircodata[,c(1,2,3,6)])
	uniqVirologic = unique(virologicdata[,c(1,2,3,6)])
	
	pvalue<-numeric(nrow(uniqVirco))
	
	## Get fisher's exact test p-value
	for (i in 1:nrow(uniqVirco)) {
		ctable=rbind(uniqVirco[i,c(4,3)], uniqVirologic[i,c(4,3)])
		ctable[,2]=ctable[,2]-ctable[,1]
		pvalue[i]=fisher.test(ctable)$p.value
	} 
	
	## Format output
	out.table = data.frame(uniqVirco[,c(1,2)], virco=paste(uniqVirco[,4],'/',uniqVirco[,3], sep=''),
		virologic=paste(uniqVirologic[,4],'/',uniqVirologic[,3], sep=''), pvalue=pvalue)
		
	write.table(out.table, file='out.txt', sep='\t', row.names=F, col.names=T, quote=F)
}