From 53877492ff97dd3fb71686986723a9186008f41c Mon Sep 17 00:00:00 2001
From: Mia Croiset <mia.croiset@ens-lyon.fr>
Date: Fri, 3 Mar 2023 09:26:10 +0100
Subject: [PATCH] ARGS hicstuff align_bowtie2 & fragment_enzyme
---
bin/hicstuff_fragments.py | 22 +++-
conf/hicstuff.config | 140 ++++++++++++----------
modules/local/hicstuff/align_bowtie2.nf | 3 +-
modules/local/hicstuff/fragment_enzyme.nf | 9 +-
4 files changed, 104 insertions(+), 70 deletions(-)
diff --git a/bin/hicstuff_fragments.py b/bin/hicstuff_fragments.py
index c49378e..489945d 100755
--- a/bin/hicstuff_fragments.py
+++ b/bin/hicstuff_fragments.py
@@ -7,10 +7,22 @@ if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("-e", "--enzyme")
parser.add_argument("-i", "--index")
+ parser.add_argument("-m", "--min_size")
+ parser.add_argument("-c", "--circular")
+ parser.add_argument("-o", "--output_contigs")
+ parser.add_argument("-f", "--output_frags")
+ parser.add_argument("-p", "--plot")
+ parser.add_argument("-g", "--fig_path")
args = parser.parse_args()
enzyme = args.enzyme
index = args.index
+ min_size = args.min_size
+ circular = args.circular
+ output_contigs = args.output_contigs
+ output_frags = args.output_frags
+ plot = args.plot
+ fig_path = args.fig_path
#hicstuff case sensitive enzymes adaptation
@@ -30,11 +42,11 @@ if __name__ == "__main__":
hcd.write_frag_info(
index,
enzyme,
- min_size=0,
- circular=False,
- output_contigs="info_contigs.txt",
- output_frags="fragments_list.txt",
+ min_size,
+ circular,
+ output_contigs,
+ output_frags
)
# Log fragment size distribution
- hcd.frag_len(frags_file_name="fragments_list.txt", plot=False, fig_path="frags_hist.pdf")
+ hcd.frag_len(output_frags, plot=plot, fig_path=fig_path)
diff --git a/conf/hicstuff.config b/conf/hicstuff.config
index 6b1178a..578ded8 100644
--- a/conf/hicstuff.config
+++ b/conf/hicstuff.config
@@ -1,64 +1,3 @@
-process {
- //Default
- publishDir = [
- path: { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" },
- mode: 'copy',
- saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
- ]
-
- withName: 'BOWTIE2_ALIGNMENT' {
- ext.prefix = { "${meta.id}_${meta.chunk}_${meta.mates}" }
- publishDir = [
- path: { "${params.outdir}/hicstuff/reads"},
- mode: 'copy'
- ]
- }
-
- withName: 'FRAGMENT_ENZYME' {
- publishDir = [
- path: { "${params.outdir}/hicstuff/fragment_enzyme" },
- mode: 'copy'
- ]
- }
-
- withName: 'BAM2PAIRS' {
- publishDir = [
- path: { "${params.outdir}/hicstuff/pairs" },
- mode: 'copy'
- ]
- }
- withName: 'BUILD_MATRIX' {
- publishDir = [
- path: { "${params.outdir}/hicstuff/matrix" },
- mode: 'copy'
- ]
- }
-
- //**********************************************
- // PREPARE_GENOME
- withName: 'BOWTIE2_BUILD' {
- publishDir = [
- path: { "${params.outdir}/genome/bowtie2_index" },
- mode: 'copy'
- ]
- }
-
- withName: 'CUSTOM_GETCHROMSIZES' {
- publishDir = [
- path: { "${params.outdir}/genome" },
- mode: 'copy'
- ]
- }
-
- withName: 'GET_RESTRICTION_FRAGMENTS' {
- publishDir = [
- path: { "${params.outdir}/genome" },
- mode: 'copy'
- ]
- }
-
-}
-
params {
// Input options
@@ -179,7 +118,86 @@ params {
max_cpus = 16
max_time = '240.h'
+ // Hicstuff
+ hicstuff_bwt2_align_opts = '--very-sensitive-local'
+ hicstuff_min_size = 0
+ hicstuff_circular = 'False'
+ hicstuff_output_contigs = 'info_contigs.txt'
+ hicstuff_output_frags = 'fragments_list.txt'
+ hicstuff_frags_plot = 'False'
+ hicstuff_frags_plot_path = 'frags_hist.pdf'
}
+
+process {
+ //Default
+ publishDir = [
+ path: { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" },
+ mode: 'copy',
+ saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+ ]
+
+ withName: 'BOWTIE2_ALIGNMENT' {
+ ext.prefix = { "${meta.id}_${meta.chunk}_${meta.mates}" }
+ ext.args = params.hicstuff_bwt2_align_opts
+ publishDir = [
+ path: { "${params.outdir}/hicstuff/reads"},
+ mode: 'copy'
+ ]
+ }
+
+ withName: 'FRAGMENT_ENZYME' {
+ ext.args = { [
+ " -m ${params.hicstuff_min_size}",
+ " -c ${params.hicstuff_circular}",
+ " -o ${params.hicstuff_output_contigs}",
+ " -f ${params.hicstuff_output_frags}",
+ " -p ${params.hicstuff_frags_plot}",
+ " -g ${params.hicstuff_frags_plot_path}"
+ ].join('').trim() }
+ publishDir = [
+ path: { "${params.outdir}/hicstuff/fragment_enzyme" },
+ mode: 'copy'
+ ]
+ }
+
+ withName: 'BAM2PAIRS' {
+ publishDir = [
+ path: { "${params.outdir}/hicstuff/pairs" },
+ mode: 'copy'
+ ]
+ }
+ withName: 'BUILD_MATRIX' {
+ publishDir = [
+ path: { "${params.outdir}/hicstuff/matrix" },
+ mode: 'copy'
+ ]
+ }
+
+ //**********************************************
+ // PREPARE_GENOME
+ withName: 'BOWTIE2_BUILD' {
+ publishDir = [
+ path: { "${params.outdir}/genome/bowtie2_index" },
+ mode: 'copy'
+ ]
+ }
+
+ withName: 'CUSTOM_GETCHROMSIZES' {
+ publishDir = [
+ path: { "${params.outdir}/genome" },
+ mode: 'copy'
+ ]
+ }
+
+ withName: 'GET_RESTRICTION_FRAGMENTS' {
+ publishDir = [
+ path: { "${params.outdir}/genome" },
+ mode: 'copy'
+ ]
+ }
+
+}
+
profiles {
debug { process.beforeScript = 'echo $HOSTNAME' }
conda {
diff --git a/modules/local/hicstuff/align_bowtie2.nf b/modules/local/hicstuff/align_bowtie2.nf
index a3f2f1c..188836d 100644
--- a/modules/local/hicstuff/align_bowtie2.nf
+++ b/modules/local/hicstuff/align_bowtie2.nf
@@ -16,13 +16,14 @@ process BOWTIE2_ALIGNMENT {
script:
def prefix = task.ext.prefix ?: "${meta.id}"
+ def args = task.ext.args ?: ''
"""
INDEX=`find -L ./ -name "*.rev.1.bt2" | sed "s/\\.rev.1.bt2\$//"`
[ -z "\$INDEX" ] && INDEX=`find -L ./ -name "*.rev.1.bt2l" | sed "s/\\.rev.1.bt2l\$//"`
[ -z "\$INDEX" ] && echo "Bowtie2 index files not found" 1>&2 && exit 1
- bowtie2 --very-sensitive-local -p $task.cpus -x \$INDEX -U $reads > ${prefix}.tmp
+ bowtie2 $args -p $task.cpus -x \$INDEX -U $reads > ${prefix}.tmp
samtools view -F 2048 -h -@ $task.cpus ${prefix}.tmp | samtools sort -n -@ $task.cpus -o ${prefix}.bam -
"""
diff --git a/modules/local/hicstuff/fragment_enzyme.nf b/modules/local/hicstuff/fragment_enzyme.nf
index dafa3cf..d17fdfe 100644
--- a/modules/local/hicstuff/fragment_enzyme.nf
+++ b/modules/local/hicstuff/fragment_enzyme.nf
@@ -10,12 +10,15 @@ process FRAGMENT_ENZYME {
tuple val(meta), path(fasta)
output:
- tuple val(meta), path("info_contigs.txt") , emit: info_contigs
- tuple val(meta), path("fragments_list.txt"), emit: fragments_list
+ tuple val(meta), path(params.hicstuff_output_contigs) , emit: info_contigs
+ tuple val(meta), path(params.hicstuff_output_frags), emit: fragments_list
script:
+
+ def args = task.ext.args ?: ''
+
"""
- hicstuff_fragments.py -e ${digestion} -i ${fasta}
+ hicstuff_fragments.py -e ${digestion} -i ${fasta} ${args}
"""
}
--
GitLab