From ff033f30f46bd1e1e6d9cf170e832fae5282b6df Mon Sep 17 00:00:00 2001
From: Xavier Grand <157-xgrand@users.noreply.gitbio.ens-lyon.fr>
Date: Mon, 25 Jul 2022 13:48:11 +0200
Subject: [PATCH] add STAR and HTseq to RNAseq_XGR.nf
---
src/RNAseq_XGR.nf | 1 +
src/nf_modules/star/main_2.7.8a.nf | 53 ++++++++++++++++++++++++++++++
2 files changed, 54 insertions(+)
diff --git a/src/RNAseq_XGR.nf b/src/RNAseq_XGR.nf
index 2992340..45b2ace 100644
--- a/src/RNAseq_XGR.nf
+++ b/src/RNAseq_XGR.nf
@@ -151,6 +151,7 @@ workflow {
Channel
.fromPath( idx_genome )
.ifEmpty { error "Cannot find idexed genome reference files" }
+ .map{it -> [(it.baseName =~ /([^\.]*)/)[0][1], it]}
.set { genome_indexed_input }
mapping_withindex(genome_indexed_input, fastp.out.fastq)
}
diff --git a/src/nf_modules/star/main_2.7.8a.nf b/src/nf_modules/star/main_2.7.8a.nf
index 44e6693..13d0af1 100644
--- a/src/nf_modules/star/main_2.7.8a.nf
+++ b/src/nf_modules/star/main_2.7.8a.nf
@@ -181,6 +181,59 @@ STAR --runThreadN ${task.cpus} \
--outSAMstrandField intronMotif \
--chimOutType WithinBAM
+mv ${reads_id}.Aligned.sortedByCoord.out.bam ${reads_id}.bam
+"""
+}
+
+
+process mapping_withindex {
+ container = "${container_url}"
+ label "big_mem_multi_cpus"
+ if (params.star_mapping_fastq_out != "") {
+ publishDir "results/${params.star_mapping_fastq_out}", mode: 'copy'
+ }
+
+ input:
+ tuple val(index_id), path(index)
+ tuple val(reads_id), path(reads)
+
+ output:
+ path "*.Log.final.out", emit: report
+ tuple val(reads_id), path("*.bam"), emit: bam
+
+ script:
+if (reads_id instanceof List){
+ file_prefix = reads_id[0]
+ } else {
+ file_prefix = reads_id
+ }
+
+if (reads.size() == 2)
+"""
+STAR --runThreadN ${task.cpus} \
+--genomeDir index/ \
+--readFilesCommand zcat \
+--readFilesIn ${reads[0]} ${reads[1]} \
+--outFileNamePrefix ${reads_id}. \
+--alignIntronMax 10000 \
+--outSAMtype BAM SortedByCoordinate \
+--outSAMstrandField intronMotif
+
+mv ${reads_id}.Aligned.sortedByCoord.out.bam ${reads_id}.bam
+"""
+else
+"""
+mkdir -p index
+mv ${index} index/
+STAR --runThreadN ${task.cpus} \
+--genomeDir index/ \
+--readFilesCommand zcat \
+--readFilesIn ${reads} \
+--outFileNamePrefix ${reads_id}. \
+--alignIntronMax 10000 \
+--outSAMtype BAM SortedByCoordinate \
+--outSAMstrandField intronMotif
+
mv ${reads_id}.Aligned.sortedByCoord.out.bam ${reads_id}.bam
"""
}
\ No newline at end of file
--
GitLab