From e288490077795590445c2200d4ec50ba0ded65fe Mon Sep 17 00:00:00 2001
From: Mia Croiset <mia.croiset@ens-lyon.fr>
Date: Fri, 3 Mar 2023 17:51:36 +0100
Subject: [PATCH] create build_matrix_cool module + finish args def
---
bin/hicstuff_bam2pairs.py | 4 +++-
bin/hicstuff_build_matrix.py | 1 +
conf/hicstuff.config | 12 +++++++++-
modules/local/hicstuff/bam2pairs.nf | 2 +-
modules/local/hicstuff/build_matrix.nf | 11 +++++----
modules/local/hicstuff/build_matrix_cool.nf | 25 +++++++++++++++++++++
6 files changed, 48 insertions(+), 7 deletions(-)
create mode 100644 modules/local/hicstuff/build_matrix_cool.nf
diff --git a/bin/hicstuff_bam2pairs.py b/bin/hicstuff_bam2pairs.py
index ea456aa..6e3cfeb 100755
--- a/bin/hicstuff_bam2pairs.py
+++ b/bin/hicstuff_bam2pairs.py
@@ -146,6 +146,7 @@ if __name__ == "__main__":
parser.add_argument("-q", "--min_qual")
parser.add_argument("-e", "--enzyme")
parser.add_argument("-f", "--fasta")
+ parser.add_argument("-c", "--circular")
args = parser.parse_args()
bam1 = args.bam1
@@ -156,6 +157,7 @@ if __name__ == "__main__":
min_qual = int(args.min_qual)
enzyme = args.enzyme
fasta = args.fasta
+ circular = args.circular
#hicstuff case sensitive enzymes adaptation
if enzyme == "hindiii":
@@ -175,7 +177,7 @@ if __name__ == "__main__":
for record in SeqIO.parse(hio.read_compressed(fasta), "fasta"):
# Get chromosome restriction table
restrict_table[record.id] = hcd.get_restriction_table(
- record.seq, enzyme, circular=False
+ record.seq, enzyme, circular=circular
)
hcd.attribute_fragments(out_pairs, out_idx, restrict_table)
diff --git a/bin/hicstuff_build_matrix.py b/bin/hicstuff_build_matrix.py
index 451849e..bd68ee0 100755
--- a/bin/hicstuff_build_matrix.py
+++ b/bin/hicstuff_build_matrix.py
@@ -5,6 +5,7 @@ import argparse
import pysam as ps
import pandas as pd
import shutil as st
+import subprocess as sp
import itertools
from hicstuff_log import logger
import hicstuff_io as hio
diff --git a/conf/hicstuff.config b/conf/hicstuff.config
index 6f53691..e670077 100644
--- a/conf/hicstuff.config
+++ b/conf/hicstuff.config
@@ -129,6 +129,7 @@ params {
hicstuff_valid_pairs = 'valid.pairs'
hicstuff_valid_idx = 'valid_idx.pairs'
hicstuff_min_qual = 30
+ hicstuff_matrix = 'abs_fragments_contacts_weighted.txt'
}
process {
@@ -167,7 +168,8 @@ process {
ext.args = { [
" -o ${params.hicstuff_valid_pairs}",
" -x ${params.hicstuff_valid_idx}",
- " -q ${params.hicstuff_min_qual}"
+ " -q ${params.hicstuff_min_qual}",
+ " -c ${params.hicstuff_circular}"
].join('').trim() }
publishDir = [
path: { "${params.outdir}/hicstuff/pairs" },
@@ -175,6 +177,14 @@ process {
]
}
withName: 'BUILD_MATRIX' {
+ ext.args = params.hicstuff_matrix
+ publishDir = [
+ path: { "${params.outdir}/hicstuff/matrix" },
+ mode: 'copy'
+ ]
+ }
+ withName: 'BUILD_MATRIX_COOL' {
+ ext.args = params.hicstuff_matrix
publishDir = [
path: { "${params.outdir}/hicstuff/matrix" },
mode: 'copy'
diff --git a/modules/local/hicstuff/bam2pairs.nf b/modules/local/hicstuff/bam2pairs.nf
index 68f7915..c8a5a14 100644
--- a/modules/local/hicstuff/bam2pairs.nf
+++ b/modules/local/hicstuff/bam2pairs.nf
@@ -1,5 +1,5 @@
process BAM2PAIRS {
- tag "$info_contigs"
+ tag "$meta1"
label 'process_high'
conda "conda-forge::python=3.9 conda-forge::biopython=1.80 conda-forge::numpy=1.22.3 conda-forge::matplotlib=3.6.3 conda-forge::pandas=1.5.3"
diff --git a/modules/local/hicstuff/build_matrix.nf b/modules/local/hicstuff/build_matrix.nf
index c81df96..d8ff6a6 100644
--- a/modules/local/hicstuff/build_matrix.nf
+++ b/modules/local/hicstuff/build_matrix.nf
@@ -1,5 +1,5 @@
process BUILD_MATRIX {
- tag "$idx_pairs"
+ tag "$meta1.id"
label 'process_single'
conda "conda-forge::python=3.9 conda-forge::biopython=1.80 conda-forge::numpy=1.22.3 conda-forge::matplotlib=3.6.3 conda-forge::pandas=1.5.3"
@@ -10,12 +10,15 @@ process BUILD_MATRIX {
tuple val(meta), path(fragments_list)
output:
- tuple val(meta), path("${meta1.id}_fragments_contacts_weighted.txt"), emit: matrix
+ tuple val(meta), path("${meta1.id}_*"), emit: matrix
script:
+
+ def args = task.ext.args ?: ''
+
"""
- hicstuff_build_matrix.py -p ${idx_pairs} -f ${fragments_list} -t graal -o abs_fragments_contacts_weighted.txt
+ hicstuff_build_matrix.py -p ${idx_pairs} -f ${fragments_list} -t graal -o ${args}
- mv abs_fragments_contacts_weighted.txt ${meta1.id}_fragments_contacts_weighted.txt
+ mv ${args} ${meta1.id}_${args}
"""
}
diff --git a/modules/local/hicstuff/build_matrix_cool.nf b/modules/local/hicstuff/build_matrix_cool.nf
new file mode 100644
index 0000000..3611061
--- /dev/null
+++ b/modules/local/hicstuff/build_matrix_cool.nf
@@ -0,0 +1,25 @@
+process BUILD_MATRIX_COOL {
+ tag "$meta1.id"
+ label 'process_single'
+
+ conda "conda-forge::python=3.9 conda-forge::biopython=1.80 conda-forge::numpy=1.22.3 conda-forge::matplotlib=3.6.3 conda-forge::pandas=1.5.3"
+ container = "lbmc/hicstuff:3.1.3"
+
+ input:
+ tuple val(meta1), path(idx_pairs)
+ tuple val(meta), path(fragments_list)
+
+ output:
+ tuple val(meta), path("${meta1.id}_*.cool"), emit: matrix
+
+ script:
+
+ def args = task.ext.args ?: ''
+ def base = args.replaceFirst(/.txt/,"")
+
+ """
+ hicstuff_build_matrix.py -p ${idx_pairs} -f ${fragments_list} -t cool -o ${args}
+
+ mv ${base}.cool ${meta1.id}_${base}.cool
+ """
+}
--
GitLab