comparison baseline/filter.r @ 4:5ffd52fc35c4 draft

Uploaded
author davidvanzessen
date Mon, 12 Dec 2016 05:22:37 -0500
parents
children
comparison
equal deleted inserted replaced
3:beaa487ecf43 4:5ffd52fc35c4
1 arg = commandArgs(TRUE)
2 summaryfile = arg[1]
3 gappedfile = arg[2]
4 selection = arg[3]
5 output = arg[4]
6 print(paste("selection = ", selection))
7
8
9 summarydat = read.table(summaryfile, header=T, sep="\t", fill=T, stringsAsFactors=F)
10 gappeddat = read.table(gappedfile, header=T, sep="\t", fill=T, stringsAsFactors=F)
11
12 #dat = data.frame(merge(gappeddat, summarydat, by="Sequence.ID", all.x=T))
13
14 dat = cbind(gappeddat, summarydat$AA.JUNCTION)
15
16 colnames(dat)[length(dat)] = "AA.JUNCTION"
17
18 dat$VGene = gsub("^Homsap ", "", dat$V.GENE.and.allele)
19 dat$VGene = gsub("[*].*", "", dat$VGene)
20
21 dat$DGene = gsub("^Homsap ", "", dat$D.GENE.and.allele)
22 dat$DGene = gsub("[*].*", "", dat$DGene)
23
24 dat$JGene = gsub("^Homsap ", "", dat$J.GENE.and.allele)
25 dat$JGene = gsub("[*].*", "", dat$JGene)
26
27 #print(str(dat))
28
29 dat$past = do.call(paste, c(dat[unlist(strsplit(selection, ","))], sep = ":"))
30
31 dat = dat[!duplicated(dat$past), ]
32
33 dat = dat[dat$Functionality != "No results" & dat$Functionality != "unproductive",]
34
35 write.table(x=dat, file=output, sep="\t",quote=F,row.names=F,col.names=T)