--- a/mutation_analysis.r	Wed Sep 10 10:33:29 2014 -0400
+++ b/mutation_analysis.r	Wed Sep 17 07:25:17 2014 -0400
@@ -9,19 +9,10 @@
 dat = read.table(input, header=T, sep="\t", fill=T, stringsAsFactors=F)
 
 datSum = read.table(summaryinput, header=T, sep="\t", fill=T, stringsAsFactors=F)
-datSum = datSum[,c("Sequence.ID", "AA.JUNCTION")]
+datSum = datSum[,c("Sequence.ID","J.GENE.and.allele", "AA.JUNCTION")]
 
 dat = merge(dat, datSum, by="Sequence.ID", all.x=T)
 
-#dat = dat[dat$Functionality == "productive",]
-
-dat$VGene = gsub("^Homsap ", "", dat$V.GENE.and.allele)
-dat$VGene = gsub("[*].*", "", dat$VGene) 
-
-dat$past = paste(dat$AA.JUNCTION, dat$VGene)
-
-#dat = dat[!duplicated(dat$past), ]
-
 if(length(dat$Sequence.ID) == 0){
 	setwd(outputdir)
 	result = data.frame(x = rep(0, 5), y = rep(0, 5), z = rep(NA, 5))
@@ -120,6 +111,15 @@
   dat[is.na(dat[,col]),] = 0
 }
 
+dat$VGene = gsub("^Homsap ", "", dat$V.GENE.and.allele)
+dat$VGene = gsub("[*].*", "", dat$VGene)
+dat$JGene = gsub("^Homsap ", "", dat$J.GENE.and.allele)
+dat$JGene = gsub("[*].*", "", dat$JGene)
+
+dat$past = paste(dat$AA.JUNCTION, dat$VGene, dat$JGene, (dat$FR1.IMGT.Nb.of.mutations + dat$CDR1.IMGT.Nb.of.mutations + dat$FR2.IMGT.Nb.of.mutations + dat$CDR2.IMGT.Nb.of.mutations + dat$FR3.IMGT.Nb.of.mutations))
+
+dat = dat[!duplicated(dat$past), ]
+
 VRegionMutations =  sum(dat$FR1.IMGT.Nb.of.mutations + 
                         dat$CDR1.IMGT.Nb.of.mutations + 
                         dat$FR2.IMGT.Nb.of.mutations +