HTSeq: update nf structure

src/docker_modules/HTSeq/0.8.0/Dockerfile

@@ -14,4 +14,5 @@ RUN apt-get update && \
     apt-get clean
 
 RUN pip3 install numpy==1.14.3
+RUN pip3 install pysam==0.15.0
 RUN pip3 install HTSeq==${HTSEQ_VERSION}

src/nf_modules/HTSeq/htseq.config

@@ -3,6 +3,9 @@ profiles {
     docker.temp = 'auto'
     docker.enabled = true
     process {
+      $sort_bam {
+        container = "samtools:1.7"
+      }
       $counting {
         container = "htseq:0.8.0"
       }
@@ -10,6 +13,9 @@ profiles {
   }
   sge {
     process{
+      $sort_bam {
+        beforeScript = "module purge; module load SAMtools/1.7"
+      }
       $trimming {
         beforeScript = "module purge; module load HTSeq/0.8.0"
       }

src/nf_modules/HTSeq/htseq.nf

-/*
-* htseq :
-* Imputs : sorted bams files
-* Imputs : gtf
-* Output : counts files
-*/
-/*                      quality trimming                                     */
-
 params.bam = "$baseDir/data/bam/*.bam"
 params.gtf = "$baseDir/data/annotation/*.gtf"
 
@@ -15,18 +7,36 @@ log.info "gtf files : ${params.gtf}"
 Channel
   .fromPath( params.bam )
   .ifEmpty { error "Cannot find any fastq files matching: ${params.bam}" }
+  .map { it -> [(it.baseName =~ /([^\.]*)/)[0][1], it]}
   .set { bam_files }
 Channel
   .fromPath( params.gtf )
   .ifEmpty { error "Cannot find any gtf file matching: ${params.gtf}" }
   .set { gtf_file }
 
+process sort_bam {
+  tag "$file_id"
+  cpus 4
+
+  input:
+    set file_id, file(bam) from bam_files
+
+  output:
+    set file_id, "*_sorted.sam" into sorted_bam_files
+
+  script:
+"""
+# sort bam by name
+samtools sort -@ ${task.cpus} -n -O SAM -o ${file_id}_sorted.sam ${bam}
+"""
+}
+
 process counting {
-  tag "$bam.baseName"
+  tag "$file_id"
   publishDir "results/quantification/", mode: 'copy'
 
   input:
-  file bam from bam_files
+  set file_id, file(bam) from sorted_bam_files
   file gtf from gtf_file
 
   output:
@@ -34,7 +44,9 @@ process counting {
 
   script:
 """
-htseq-count -r pos --mode=intersection-nonempty -a 10 -s no -t exon -i gene_id \
---format=bam ${bam} ${gtf} > ${bam.baseName}.count
+htseq-count ${bam} ${gtf} \
+-r pos --mode=intersection-nonempty -a 10 -s no -t exon -i gene_id \
+> ${file_id}.count
 """
 }
+

src/nf_modules/HTSeq/tests/tests.sh → src/nf_modules/HTSeq/tests.sh

-nextflow src/nf_modules/HTSeq/tests/counting.nf \
+nextflow src/nf_modules/HTSeq/htseq.nf \
   -c src/nf_modules/HTSeq/htseq.config \
   -profile docker \
   --gtf "data/tiny_dataset/annot/tiny.gff" \

src/nf_modules/HTSeq/tests/counting.nf

-params.bam = "$baseDir/data/bam/*.bam"
-params.gtf = "$baseDir/data/annotation/*.gtf"
-
-log.info "bam files : ${params.bam}"
-log.info "gtf files : ${params.gtf}"
-
-Channel
-  .fromPath( params.bam )
-  .ifEmpty { error "Cannot find any fastq files matching: ${params.bam}" }
-  .set { bam_files }
-Channel
-  .fromPath( params.gtf )
-  .ifEmpty { error "Cannot find any gtf file matching: ${params.gtf}" }
-  .set { gtf_file }
-
-process counting {
-  tag "$bam.baseName"
-  publishDir "results/quantification/", mode: 'copy'
-
-  input:
-  file bam from bam_files
-  file gtf from gtf_file
-
-  output:
-  file "*.count" into count_files
-
-  script:
-"""
-htseq-count -r pos --mode=intersection-nonempty -a 10 -s no -t exon -i gene_id \
---format=bam ${bam} ${gtf} > ${bam.baseName}.count
-"""
-}
-