annotate tmp/igat.r @ 98:5ffbf40cdd4b draft

Uploaded
author davidvanzessen
date Thu, 16 Jun 2016 05:05:47 -0400
parents 6e8dfbe164c6
children 86206431cbb0
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
95
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
1 args <- commandArgs(trailingOnly = TRUE)
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
2
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
3 imgt.dir = args[1]
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
4 merged.file = args[2]
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
5
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
6 merged = read.table(merged.file, header=T, sep="\t", fill=T, stringsAsFactors=F)
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
7
97
6e8dfbe164c6 Uploaded
davidvanzessen
parents: 95
diff changeset
8 merged = merged[!grepl("unmatched", merged$best_match),]
6e8dfbe164c6 Uploaded
davidvanzessen
parents: 95
diff changeset
9
95
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
10 for(f in list.files(imgt.dir, pattern="*.txt$")){
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
11 print(paste("filtering", f))
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
12 path = paste(imgt.dir, f, sep="")
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
13 dat = read.table(path, header=T, sep="\t", fill=T, quote="", stringsAsFactors=F)
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
14
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
15 dat = dat[dat$Sequence.ID %in% merged$Sequence.ID,]
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
16
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
17 if("FR1.IMGT" %in% colnames(dat)){
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
18 dat$FR1.IMGT = ""
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
19 }
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
20
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
21 write.table(dat, path, quote=F, sep="\t", row.names=F, col.names=T)
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
22 }