95
|
1 args <- commandArgs(trailingOnly = TRUE)
|
|
2
|
|
3 imgt.dir = args[1]
|
|
4 merged.file = args[2]
|
99
|
5 gene = args[3]
|
95
|
6
|
|
7 merged = read.table(merged.file, header=T, sep="\t", fill=T, stringsAsFactors=F)
|
|
8
|
99
|
9 if(gene != "-"){
|
|
10 merged = merged[grepl(gene, merged$best_match),]
|
|
11 }
|
|
12
|
97
|
13 merged = merged[!grepl("unmatched", merged$best_match),]
|
|
14
|
95
|
15 for(f in list.files(imgt.dir, pattern="*.txt$")){
|
|
16 print(paste("filtering", f))
|
|
17 path = paste(imgt.dir, f, sep="")
|
|
18 dat = read.table(path, header=T, sep="\t", fill=T, quote="", stringsAsFactors=F)
|
|
19
|
|
20 dat = dat[dat$Sequence.ID %in% merged$Sequence.ID,]
|
|
21
|
99
|
22 if(nrow(dat) > 0 & "FR1.IMGT" %in% colnames(dat)){
|
95
|
23 dat$FR1.IMGT = ""
|
|
24 }
|
|
25
|
|
26 write.table(dat, path, quote=F, sep="\t", row.names=F, col.names=T)
|
|
27 }
|