From 93338e8dbb3b6ea01697455dfbf7262c7300df75 Mon Sep 17 00:00:00 2001
From: Laurent Modolo <laurent.modolo@ens-lyon.fr>
Date: Thu, 20 Apr 2023 15:38:54 +0200
Subject: [PATCH] add split process
---
src/main.nf | 4 +++-
src/modules/fastkmers.nf | 2 +-
src/modules/split.nf | 23 +++++++++++++++++++++++
3 files changed, 27 insertions(+), 2 deletions(-)
create mode 100644 src/modules/split.nf
diff --git a/src/main.nf b/src/main.nf
index 7269cae..5034df4 100644
--- a/src/main.nf
+++ b/src/main.nf
@@ -5,6 +5,7 @@ Testing pipeline for marseq scRNASeq analysis
*/
include { FASTKMERS } from "./modules/fastkmers"
+include { SPLIT } from "./modules/split"
Channel.fromPath( file(params.csv) )
@@ -18,5 +19,6 @@ Channel.fromPath( file(params.csv) )
.set{ fastq }
workflow {
- FASTKMERS(fastq.r1.mix(fastq.r2).splitFastq( by: 1000000, compress: true, file: true ))
+ SPLIT(fastq.r1.mix(fastq.r2))
+ FASTKMERS(SPLIT.out.fastq.transpose())
}
diff --git a/src/modules/fastkmers.nf b/src/modules/fastkmers.nf
index 4384f0b..ebbdf45 100644
--- a/src/modules/fastkmers.nf
+++ b/src/modules/fastkmers.nf
@@ -3,7 +3,7 @@
*/
process FASTKMERS {
tag "$meta.id"
- label 'process_single'
+ label 'big_mem_mono_cpus'
publishDir "results/${meta.specie}/${meta.sex}/${meta.read}/", mode: 'copy'
container "lbmc/fastkmers:025efdf"
diff --git a/src/modules/split.nf b/src/modules/split.nf
new file mode 100644
index 0000000..7324914
--- /dev/null
+++ b/src/modules/split.nf
@@ -0,0 +1,23 @@
+process SPLIT {
+ tag "$meta.id"
+ label 'small_mem_mono_cpus'
+
+ container "alpine:3.17"
+
+ input:
+ tuple val(meta), path(fastq)
+
+ output:
+ tuple val(meta), path("*.fastq.gz"), emit: fastq
+ path "versions.yml" , emit: versions
+
+ script:
+ def args = task.ext.args ?: ''
+ """
+ zcat ${fastq} | split -d -l4000000 --filter='gzip > \$FILE.fastq.gz' ${fastq.simpleName}_
+
+ "${task.process}":
+ split: v1.35.0
+ END_VERSIONS
+ """
+}
--
GitLab