|
0
|
1 args <- commandArgs(trailingOnly = TRUE)
|
|
|
2
|
|
|
3 input=args[1]
|
|
|
4 column=as.numeric(args[2])
|
|
|
5 header=(args[3] == "yes")
|
|
2
|
6 regex_filter=args[4]
|
|
|
7 out_file=args[5]
|
|
|
8
|
|
|
9 print(regex_filter)
|
|
0
|
10
|
|
1
|
11 dat = read.table(input, header=header, sep="\t", fill=T, stringsAsFactors=F, quote="")
|
|
0
|
12
|
|
2
|
13 dat.names = names(dat)
|
|
0
|
14
|
|
2
|
15 dat$filtered = gsub("\\(.*", "", dat[,column])
|
|
|
16
|
|
|
17 duplicates = dat[duplicated(dat$filtered),"filtered"]
|
|
0
|
18
|
|
2
|
19 dat = dat[dat[,"filtered"] %in% duplicates,]
|
|
0
|
20
|
|
2
|
21 dat = dat[order(dat[,"filtered"]),]
|
|
|
22
|
|
|
23 write.table(dat[,dat.names], out_file, sep="\t", row.names=F, col.names=header, quote=F)
|
