From e54720be08d3a4998a392e196c665d67a9347c9d Mon Sep 17 00:00:00 2001
From: Laurent Modolo <laurent.modolo@ens-lyon.fr>
Date: Wed, 5 May 2021 14:36:41 +0200
Subject: [PATCH] alntools: update transcript lengths computation
---
src/nf_modules/alntools/main.nf | 15 +++++++--------
1 file changed, 7 insertions(+), 8 deletions(-)
diff --git a/src/nf_modules/alntools/main.nf b/src/nf_modules/alntools/main.nf
index 8813a050..ef1e879e 100644
--- a/src/nf_modules/alntools/main.nf
+++ b/src/nf_modules/alntools/main.nf
@@ -21,17 +21,16 @@ process bam2ec {
script:
"""
-awk -F"\t" '
-\$3=="exon"
- {
- ID=substr(\$9, length(\$9)-16, 15);
- L[ID]+=\$5-\$4+1
+awk -F"[\t;]" '
+\$3=="exon" {
+ ID=gensub(/transcript_id \"(.*)\"/, "\\1", "g", \$11);
+ LEN[ID]+=\$5-\$4+1;
}
END{
- for(i in L)
- {print i"\t"L[i]}
+ for(i in LEN)
+ {print i"\t"LEN[i]}
}
-' ${gtf} ${gtf.simpleName}_transcripts_lengths.tsv
+' ${gtf} > ${gtf.simpleName}_transcripts_lengths.tsv
alntools bam2ec ${params.bam2sec} -t ${gtf.simpleName}_transcripts_lengths.tsv -c 8 ${bam} ${bam.simpleName}.bin
"""
}
\ No newline at end of file
--
GitLab