view new_imgt.r @ 16:4e596473c25c draft

Uploaded
author davidvanzessen
date Wed, 23 Nov 2016 09:05:40 -0500
parents c33d93683a09
children fe44a905aee9
line wrap: on
line source

args <- commandArgs(trailingOnly = TRUE)

imgt.dir = args[1]
merged.file = args[2]
gene = args[3]

merged = read.table(merged.file, header=T, sep="\t", fill=T, stringsAsFactors=F)

if(gene != "-"){
	merged = merged[grepl(paste("^", gene, sep=""), merged$best_match),]
} else {
	merged = merged[!grepl("unmatched", merged$best_match),]
}

merged = merged[!grepl("unmatched", merged$best_match),]

for(f in list.files(imgt.dir, pattern="*.txt$")){
	#print(paste("filtering", f))
	path = paste(imgt.dir, f, sep="")
	dat = read.table(path, header=T, sep="\t", fill=T, quote="", stringsAsFactors=F, check.names=FALSE)
	
	dat = dat[dat[,"Sequence ID"] %in% merged$Sequence.ID,]
	
	if(nrow(dat) > 0 & grepl("^8_", f)){ #change the FR1 columns to 0 in the "8_..." file
		dat[,grepl("^FR1", names(dat))] = 0
	}
	
	write.table(dat, path, quote=F, sep="\t", row.names=F, col.names=T, na="")
}