# HG changeset patch
# User davidvanzessen
# Date 1428400667 14400
# Node ID 31eee1b3d7df6fd138dde0c7cfdef25915819e0f
# Parent  28b8d980db22c46b664e5e9497bddb3c1971d8ca
Uploaded

diff -r 28b8d980db22 -r 31eee1b3d7df mutation_analysis.r
--- a/mutation_analysis.r	Tue Apr 07 04:46:35 2015 -0400
+++ b/mutation_analysis.r	Tue Apr 07 05:57:47 2015 -0400
@@ -130,20 +130,27 @@
 totalMutationsAtGC_columns = paste(rep(regions, each=6), c(".IMGT.g.a",".IMGT.c.t",".IMGT.c.a",".IMGT.g.c",".IMGT.c.g",".IMGT.g.t"), sep="")
 dat$totalMutationsAtGC = apply(dat, FUN=sum_by_row, 1, columns=totalMutationsAtGC_columns)
 
-silentMutations_columns = paste(regions, ".IMGT.Nb.of.silent.mutations", sep="")
-silentMutations_columns
-dat[,silentMutations_columns]
-dat$silentMutations = apply(dat, FUN=sum_by_row, 1, columns=silentMutations_columns)
+FRRegions = regions[grepl("FR", regions)]
+CDRRegions = regions[grepl("CDR", regions)]
+
+FR_silentMutations_columns = paste(FRRegions, ".IMGT.Nb.of.silent.mutations", sep="")
+dat$silentMutationsFR = apply(dat, FUN=sum_by_row, 1, columns=FR_silentMutations_columns)
 
-nonSilentMutations_columns = paste(regions, ".IMGT.Nb.of.nonsilent.mutations", sep="")
-dat$nonSilentMutations = apply(dat, FUN=sum_by_row, 1, columns=nonSilentMutations_columns)
+CDR_silentMutations_columns = paste(CDRRegions, ".IMGT.Nb.of.silent.mutations", sep="")
+dat$silentMutationsCDR = apply(dat, FUN=sum_by_row, 1, columns=CDR_silentMutations_columns)
+
+FR_nonSilentMutations_columns = paste(FRRegions, ".IMGT.Nb.of.nonsilent.mutations", sep="")
+dat$nonSilentMutationsFR = apply(dat, FUN=sum_by_row, 1, columns=FR_nonSilentMutations_columns)
+
+CDR_nonSilentMutations_columns = paste(CDRRegions, ".IMGT.Nb.of.nonsilent.mutations", sep="")
+dat$nonSilentMutationsCDR = apply(dat, FUN=sum_by_row, 1, columns=CDR_nonSilentMutations_columns)
 
 
 setwd(outputdir)
 
 nts = c("a", "t", "g", "c")
 zeros=rep(0, 4)
-matrx = matrix(data = 0, ncol=((length(genes) + 1) * 3),nrow=6)
+matrx = matrix(data = 0, ncol=((length(genes) + 1) * 3),nrow=7)
 for(i in 1:length(genes)){
   gene = genes[i]
   tmp = dat[grepl(paste(".*", gene, ".*", sep=""), dat$best_match),]
@@ -169,9 +176,12 @@
   matrx[5,x] = sum(tmp$totalMutationsAtGC)
   matrx[5,y] = sum(tmp$VRegionMutations)
   matrx[5,z] = round(matrx[5,x] / matrx[5,y] * 100, digits=1)
-  matrx[6,x] = sum(tmp$silentMutations)
-  matrx[6,y] = sum(tmp$nonSilentMutations)
+  matrx[6,x] = sum(tmp$silentMutationsFR)
+  matrx[6,y] = sum(tmp$nonSilentMutationsFR)
   matrx[6,z] = round(matrx[6,x] / matrx[6,y] * 100, digits=1)
+  matrx[7,x] = sum(tmp$silentMutationsCDR)
+  matrx[7,y] = sum(tmp$nonSilentMutationsCDR)
+  matrx[7,z] = round(matrx[7,x] / matrx[7,y] * 100, digits=1)
   
   
   transitionTable = data.frame(A=zeros,C=zeros,G=zeros,T=zeros)
@@ -232,9 +242,12 @@
 matrx[5,x] = sum(tmp$totalMutationsAtGC)
 matrx[5,y] = sum(tmp$VRegionMutations)
 matrx[5,z] = round(matrx[5,x] / matrx[5,y] * 100, digits=1)
-matrx[6,x] = sum(tmp$silentMutations)
-matrx[6,y] = sum(tmp$nonSilentMutations)
+matrx[6,x] = sum(tmp$silentMutationsFR)
+matrx[6,y] = sum(tmp$nonSilentMutationsFR)
 matrx[6,z] = round(matrx[6,x] / matrx[6,y] * 100, digits=1)
+matrx[7,x] = sum(tmp$silentMutationsCDR)
+matrx[7,y] = sum(tmp$nonSilentMutationsCDR)
+matrx[7,z] = round(matrx[7,x] / matrx[7,y] * 100, digits=1)
 
 transitionTable = data.frame(A=1:4,C=1:4,G=1:4,T=1:4)
 row.names(transitionTable) = c("A", "C", "G", "T")
@@ -271,7 +284,7 @@
 
 
 result = data.frame(matrx)
-row.names(result) = c("Number of Mutations (%)", "Transition (%)", "Transversions (%)", "Transitions at G C (%)", "Targeting of C.G (%)", "Silent/Non Silent (%)")
+row.names(result) = c("Number of Mutations (%)", "Transition (%)", "Transversions (%)", "Transitions at G C (%)", "Targeting of C.G (%)", "FR S/R (%)", "CDR S/R (%)")
 
 write.table(x=result, file="mutations.txt", sep=",",quote=F,row.names=T,col.names=F)