comparison clean_rm_output.R @ 1:814cba36e435 draft

Uploaded
author mvdbeek
date Mon, 21 Feb 2022 10:21:39 +0000
parents ea6a3059a6af
children 7f1032da7a0a
comparison
equal deleted inserted replaced
0:ea6a3059a6af 1:814cba36e435
1 #!/usr/bin/env Rscript 1 #!/usr/bin/env Rscript
2 suppressPackageStartupMessages(library(rtracklayer)) 2 suppressPackageStartupMessages(library(rtracklayer))
3 suppressPackageStartupMessages(library(parallel))
4
3 5
4 gff_cleanup = function(gff){ 6 gff_cleanup = function(gff){
5 ## remove overlapin annotation track - assign new annot 7 ## remove overlapin annotation track - assign new annot
6 gff_disjoin = disjoin(gff, with.revmap=TRUE) 8 gff_disjoin = disjoin(gff, with.revmap=TRUE)
7 ## append annotation: 9 ## append annotation:
48 infile = commandArgs(T)[1] 50 infile = commandArgs(T)[1]
49 outfile = commandArgs(T)[2] 51 outfile = commandArgs(T)[2]
50 52
51 ## infile = "./test_data/raw_rm.out" 53 ## infile = "./test_data/raw_rm.out"
52 54
53 rm_out = read.table(infile, as.is=TRUE, sep="", skip = 2, fill=TRUE, header=FALSE) 55 rm_out = read.table(infile, as.is=TRUE, sep="", skip = 2, fill=TRUE, header=FALSE, col.names=paste0("V",1:16))
54 56
55 gff = GRanges(seqnames = rm_out$V5, ranges = IRanges(start = rm_out$V6, end=rm_out$V7)) 57 gff = GRanges(seqnames = rm_out$V5, ranges = IRanges(start = rm_out$V6, end=rm_out$V7))
56 58
57 # repeat class after # symbol - syntax 1 59 # repeat class after # symbol - syntax 1
58 gff$Name=rm_out$V11 60 gff$Name=rm_out$V11