From 72ffaf953089c0259de9157fee10f690643ce469 Mon Sep 17 00:00:00 2001
From: Laurent Modolo <laurent@modolo.fr>
Date: Wed, 20 Mar 2019 18:19:14 +0100
Subject: [PATCH] macs2: add nf module file
---
src/nf_modules/macs2/peak_calling.config | 36 ++++++++++++++++++
src/nf_modules/macs2/peak_calling.nf | 47 ++++++++++++++++++++++++
src/nf_modules/macs2/tests.sh | 8 ++++
3 files changed, 91 insertions(+)
create mode 100644 src/nf_modules/macs2/peak_calling.config
create mode 100644 src/nf_modules/macs2/peak_calling.nf
create mode 100755 src/nf_modules/macs2/tests.sh
diff --git a/src/nf_modules/macs2/peak_calling.config b/src/nf_modules/macs2/peak_calling.config
new file mode 100644
index 00000000..fa0c6d38
--- /dev/null
+++ b/src/nf_modules/macs2/peak_calling.config
@@ -0,0 +1,36 @@
+profiles {
+ docker {
+ docker.temp = 'auto'
+ docker.enabled = true
+ process {
+ withName: peak_calling {
+ container = "macs2:2.1.2"
+ cpus = 4
+ }
+ }
+ }
+ singularity {
+ singularity.enabled = true
+ process {
+ withName: peak_calling {
+ container = "file://bin/macs2:2.1.2.sif"
+ cpus = 4
+ }
+ }
+ }
+ psmn {
+ process{
+ withName: peak_calling {
+ beforeScript = "source /usr/share/lmod/lmod/init/bash; module use ~/privatemodules"
+ module = "macs2/2.1.2"
+ executor = "sge"
+ clusterOptions = "-m e -cwd -V"
+ cpus = 16
+ memory = "30GB"
+ time = "24h"
+ queue = 'E5-2670deb128A,E5-2670deb128B,E5-2670deb128C,E5-2670deb128D,E5-2670deb128E,E5-2670deb128F'
+ penv = 'openmp16'
+ }
+ }
+ }
+}
diff --git a/src/nf_modules/macs2/peak_calling.nf b/src/nf_modules/macs2/peak_calling.nf
new file mode 100644
index 00000000..1bd6a95c
--- /dev/null
+++ b/src/nf_modules/macs2/peak_calling.nf
@@ -0,0 +1,47 @@
+params.genome_size = "hs"
+params.control_tag = "control"
+log.info "bam files : ${params.bam}"
+log.info "genome size : ${params.genome_size}"
+
+Channel
+ .fromPath( params.bam )
+ .ifEmpty { error "Cannot find any bam files matching: ${params.bam}" }
+ .map { it -> [(it.baseName =~ /([^\.]*)/)[0][1], it]}
+ .set { bam_files }
+
+/* split bam Channel into control and ip if "control" keyword is detected*/
+bam_files_control = Channel.create()
+bam_files_ip = Channel.create()
+bam_files.choice(
+ bam_files_control,
+ bam_files_ip ) { a -> a[0] =~ /.*${params.control_tag}.*/ ? 0 : 1 }
+
+process peak_calling {
+ tag "${file_id}"
+ publishDir "results/peak_calling/${file_id}", mode: 'copy'
+
+ input:
+ set file_id, file(file_ip) from bam_files_ip
+ set file_id_control, file(file_control) from bam_files_control.collect()
+
+ output:
+ file "*" into peak_output
+ file "*_report.txt" into peak_calling_report
+
+ script:
+/* remove --nomodel option for real dataset */
+"""
+macs2 callpeak \
+ --nomodel \
+ --treatment ${file_ip} \
+ --control ${file_control} \
+ --name ${file_id} \
+ --gsize ${params.genome_size} 2> \
+${file_ip}_macs2_report.txt
+
+if grep -q "ERROR" ${file_ip}_macs2_report.txt; then
+ echo "MACS2 error"
+ exit 1
+fi
+"""
+}
diff --git a/src/nf_modules/macs2/tests.sh b/src/nf_modules/macs2/tests.sh
new file mode 100755
index 00000000..1491b074
--- /dev/null
+++ b/src/nf_modules/macs2/tests.sh
@@ -0,0 +1,8 @@
+cp data/tiny_dataset/map/tiny_v2.bam data/tiny_dataset/map/tiny_v2_control.bam
+./nextflow src/nf_modules/macs2/peak_calling.nf \
+ -c src/nf_modules/macs2/peak_calling.config \
+ -profile docker \
+ -resume \
+ --bam "data/tiny_dataset/map/tiny_v2*.bam" \
+ --genome_size 129984 \
+ --control_tag "control"
--
GitLab