From 789f692681ef915a83873b0d88f10e564ba6bc80 Mon Sep 17 00:00:00 2001
From: Laurent Modolo <laurent.modolo@ens-lyon.fr>
Date: Thu, 29 Apr 2021 09:26:08 +0200
Subject: [PATCH] alntools: add bam2ec process
---
src/nf_modules/alntools/main.nf | 32 ++++++++++++++++++++++++++++++++
1 file changed, 32 insertions(+)
create mode 100644 src/nf_modules/alntools/main.nf
diff --git a/src/nf_modules/alntools/main.nf b/src/nf_modules/alntools/main.nf
new file mode 100644
index 00000000..e6f990d0
--- /dev/null
+++ b/src/nf_modules/alntools/main.nf
@@ -0,0 +1,32 @@
+version = "dd69682"
+container_url = "lbmc/alntools:${version}"
+
+process bam2ec {
+ container = "${container_url}"
+ label "big_mem_mono_cpus"
+ tag "$file_id"
+
+ input:
+ tuple val(file_id), path(bam)
+ tuple val(gtf_id), path(gtf)
+
+ output:
+ tuple val(file_id), path("${bam.simpleName}.bin"), emit: bin
+ tuple val(gtf_id), path("${gtf.simpleName}_transcripts_lengths.tsv"), emit: tsv
+
+ script:
+"""
+awk -F"\t" '
+$3=="exon"
+ {
+ ID=substr($9, length($9)-16, 15);
+ L[ID]+=$5-$4+1
+ }
+END{
+ for(i in L)
+ {print i"\t"L[i]}
+ }
+' ${gtf} ${gtf.simpleName}_transcripts_lengths.tsv
+alntools bam2ec -t ${gtf.simpleName}_transcripts_lengths.tsv -c 8 ${bam} ${bam.simpleName}.bin
+"""
+}
\ No newline at end of file
--
GitLab