annotate tmp/igat.r @ 99:86206431cbb0 draft

Uploaded
author davidvanzessen
date Thu, 16 Jun 2016 10:01:54 -0400
parents 6e8dfbe164c6
children 3cffb8a38bb1
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]
99
86206431cbb0 Uploaded
davidvanzessen
parents: 97
diff changeset
5 gene = args[3]
95
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
6
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
7 merged = read.table(merged.file, header=T, sep="\t", fill=T, stringsAsFactors=F)
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
8
99
86206431cbb0 Uploaded
davidvanzessen
parents: 97
diff changeset
9 if(gene != "-"){
86206431cbb0 Uploaded
davidvanzessen
parents: 97
diff changeset
10 merged = merged[grepl(gene, merged$best_match),]
86206431cbb0 Uploaded
davidvanzessen
parents: 97
diff changeset
11 }
86206431cbb0 Uploaded
davidvanzessen
parents: 97
diff changeset
12
97
6e8dfbe164c6 Uploaded
davidvanzessen
parents: 95
diff changeset
13 merged = merged[!grepl("unmatched", merged$best_match),]
6e8dfbe164c6 Uploaded
davidvanzessen
parents: 95
diff changeset
14
95
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
15 for(f in list.files(imgt.dir, pattern="*.txt$")){
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
16 print(paste("filtering", f))
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
17 path = paste(imgt.dir, f, sep="")
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
18 dat = read.table(path, header=T, sep="\t", fill=T, quote="", stringsAsFactors=F)
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
19
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
20 dat = dat[dat$Sequence.ID %in% merged$Sequence.ID,]
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
21
99
86206431cbb0 Uploaded
davidvanzessen
parents: 97
diff changeset
22 if(nrow(dat) > 0 & "FR1.IMGT" %in% colnames(dat)){
95
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
23 dat$FR1.IMGT = ""
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
24 }
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
25
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
26 write.table(dat, path, quote=F, sep="\t", row.names=F, col.names=T)
a66eb1c5374c Uploaded
davidvanzessen
parents:
diff changeset
27 }