comparison new_imgt.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 args <- commandArgs(trailingOnly = TRUE)
2
3 imgt.dir = args[1]
4 merged.file = args[2]
5 gene = args[3]
6
7 merged = read.table(merged.file, header=T, sep="\t", fill=T, stringsAsFactors=F)
8
9 if(gene != "-"){
10 merged = merged[grepl(paste("^", gene, sep=""), merged$best_match),]
11 } else {
12 merged = merged[!grepl("unmatched", merged$best_match),]
13 }
14
15 merged = merged[!grepl("unmatched", merged$best_match),]
16
17 for(f in list.files(imgt.dir, pattern="*.txt$")){
18 #print(paste("filtering", f))
19 path = paste(imgt.dir, f, sep="")
20 dat = read.table(path, header=T, sep="\t", fill=T, quote="", stringsAsFactors=F, check.names=FALSE)
21
22 dat = dat[dat[,"Sequence ID"] %in% merged$Sequence.ID,]
23
24 if(nrow(dat) > 0 & grepl("^8_", f)){ #change the FR1 columns to 0 in the "8_..." file
25 dat[,grepl("^FR1", names(dat))] = 0
26 }
27
28 write.table(dat, path, quote=F, sep="\t", row.names=F, col.names=T, na="")
29 }