Add new boolean sort_bam as input to be able to chose between sort and view in bwamem and bwamem2mem (#1125)

* feat: view is now in args2 so we can use sort

* forgot one split_cpus

* feat: update with new logic

* fix: add more info

* fix: remove split_cpus logic

modules/bwa/mem/main.nf

@@ -10,6 +10,7 @@ process BWA_MEM {
     input:
     tuple val(meta), path(reads)
     path  index
+    val   sort_bam
 
     output:
     tuple val(meta), path("*.bam"), emit: bam
@@ -20,6 +21,7 @@ process BWA_MEM {
     def args2 = task.ext.args2 ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
     def read_group = meta.read_group ? "-R ${meta.read_group}" : ""
+    def samtools_command = sort_bam ? 'sort' : 'view'
     """
     INDEX=`find -L ./ -name "*.amb" | sed 's/.amb//'`
 
@@ -29,7 +31,7 @@ process BWA_MEM {
         -t $task.cpus \\
         \$INDEX \\
         $reads \\
-        | samtools view $args2 -@ $task.cpus -bhS -o ${prefix}.bam -
+        | samtools $samtools_command $args2 --threads $task.cpus -o ${prefix}.bam -
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":

modules/bwa/mem/meta.yml

@@ -32,6 +32,10 @@ input:
         type: file
         description: BWA genome index files
         pattern: "Directory containing BWA index *.{amb,ann,bwt,pac,sa}"
+    - sort_bam:
+        type: boolean
+        description: use samtools sort (true) or samtools view (false)
+        pattern: "true or false"
 output:
     - bam:
         type: file

modules/bwamem2/mem/main.nf

@@ -10,6 +10,7 @@ process BWAMEM2_MEM {
     input:
     tuple val(meta), path(reads)
     path  index
+    val   sort_bam
 
     output:
     tuple val(meta), path("*.bam"), emit: bam
@@ -20,6 +21,7 @@ process BWAMEM2_MEM {
     def args2 = task.ext.args2 ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
     def read_group = meta.read_group ? "-R ${meta.read_group}" : ""
+    def samtools_command = sort_bam ? 'sort' : 'view'
     """
     INDEX=`find -L ./ -name "*.amb" | sed 's/.amb//'`
 
@@ -30,7 +32,7 @@ process BWAMEM2_MEM {
         -t $task.cpus \\
         \$INDEX \\
         $reads \\
-        | samtools view $args2 -@ $task.cpus -bhS -o ${prefix}.bam -
+        | samtools $samtools_command $args2 -@ $task.cpus -o ${prefix}.bam -
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":

modules/bwamem2/mem/meta.yml

@@ -11,9 +11,9 @@ keywords:
 tools:
     - bwa:
         description: |
-            BWA is a software package for mapping DNA sequences against
+            BWA-mem2 is a software package for mapping DNA sequences against
             a large reference genome, such as the human genome.
-        homepage: http://bio-bwa.sourceforge.net/
+        homepage: https://github.com/bwa-mem2/bwa-mem2
         documentation: http://www.htslib.org/doc/samtools.html
         arxiv: arXiv:1303.3997
         licence: ['MIT']
@@ -31,7 +31,11 @@ input:
     - index:
         type: file
         description: BWA genome index files
-        pattern: "Directory containing BWA index *.{amb,ann,bwt,pac,sa}"
+        pattern: "Directory containing BWA index *.{0132,amb,ann,bwt.2bit.64,pac}"
+    - sort_bam:
+        type: boolean
+        description: use samtools sort (true) or samtools view (false)
+        pattern: "true or false"
 output:
     - bam:
         type: file

tests/modules/bwa/mem/main.nf

@@ -18,7 +18,23 @@ workflow test_bwa_mem_single_end {
     fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
 
     BWA_INDEX ( fasta )
-    BWA_MEM ( input, BWA_INDEX.out.index )
+    BWA_MEM ( input, BWA_INDEX.out.index, false )
+}
+
+//
+// Test with single-end data and sort
+//
+workflow test_bwa_mem_single_end_sort {
+    input = [
+        [ id:'test', single_end:true ], // meta map
+        [
+            file(params.test_data['sarscov2']['illumina']['test_1_fastq_gz'], checkIfExists: true)
+        ]
+    ]
+    fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+
+    BWA_INDEX ( fasta )
+    BWA_MEM ( input, BWA_INDEX.out.index, true )
 }
 
 //
@@ -35,5 +51,22 @@ workflow test_bwa_mem_paired_end {
     fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
 
     BWA_INDEX ( fasta )
-    BWA_MEM ( input, BWA_INDEX.out.index )
+    BWA_MEM ( input, BWA_INDEX.out.index, false )
+}
+
+//
+// Test with paired-end data and sort
+//
+workflow test_bwa_mem_paired_end_sort {
+    input = [
+        [ id:'test', single_end:false ], // meta map
+        [
+            file(params.test_data['sarscov2']['illumina']['test_1_fastq_gz'], checkIfExists: true),
+            file(params.test_data['sarscov2']['illumina']['test_2_fastq_gz'], checkIfExists: true)
+        ]
+    ]
+    fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+
+    BWA_INDEX ( fasta )
+    BWA_MEM ( input, BWA_INDEX.out.index, true )
 }

tests/modules/bwa/mem/nextflow.config

@@ -2,4 +2,8 @@ process {
 
     publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
 
+    withName: BWA_MEM {
+        ext.args2 = { sort_bam ? "" : "-bh" }
+    }
+
 }

tests/modules/bwa/mem/test.yml

@@ -16,6 +16,24 @@
     - path: ./output/bwa/bwa/genome.sa
       md5sum: ab3952cabf026b48cd3eb5bccbb636d1
 
+- name: bwa mem single-end sort
+  command: nextflow run ./tests/modules/bwa/mem -entry test_bwa_mem_single_end_sort -c ./tests/config/nextflow.config -c ./tests/modules/bwa/mem/nextflow.config
+  tags:
+    - bwa
+    - bwa/mem
+  files:
+    - path: ./output/bwa/test.bam
+    - path: ./output/bwa/bwa/genome.bwt
+      md5sum: 0469c30a1e239dd08f68afe66fde99da
+    - path: ./output/bwa/bwa/genome.amb
+      md5sum: 3a68b8b2287e07dd3f5f95f4344ba76e
+    - path: ./output/bwa/bwa/genome.ann
+      md5sum: c32e11f6c859f166c7525a9c1d583567
+    - path: ./output/bwa/bwa/genome.pac
+      md5sum: 983e3d2cd6f36e2546e6d25a0da78d66
+    - path: ./output/bwa/bwa/genome.sa
+      md5sum: ab3952cabf026b48cd3eb5bccbb636d1
+
 - name: bwa mem paired-end
   command: nextflow run ./tests/modules/bwa/mem -entry test_bwa_mem_paired_end -c ./tests/config/nextflow.config -c ./tests/modules/bwa/mem/nextflow.config
   tags:
@@ -33,3 +51,21 @@
       md5sum: 983e3d2cd6f36e2546e6d25a0da78d66
     - path: ./output/bwa/bwa/genome.sa
       md5sum: ab3952cabf026b48cd3eb5bccbb636d1
+
+- name: bwa mem paired-end sort
+  command: nextflow run ./tests/modules/bwa/mem -entry test_bwa_mem_paired_end_sort -c ./tests/config/nextflow.config -c ./tests/modules/bwa/mem/nextflow.config
+  tags:
+    - bwa
+    - bwa/mem
+  files:
+    - path: ./output/bwa/test.bam
+    - path: ./output/bwa/bwa/genome.bwt
+      md5sum: 0469c30a1e239dd08f68afe66fde99da
+    - path: ./output/bwa/bwa/genome.amb
+      md5sum: 3a68b8b2287e07dd3f5f95f4344ba76e
+    - path: ./output/bwa/bwa/genome.ann
+      md5sum: c32e11f6c859f166c7525a9c1d583567
+    - path: ./output/bwa/bwa/genome.pac
+      md5sum: 983e3d2cd6f36e2546e6d25a0da78d66
+    - path: ./output/bwa/bwa/genome.sa
+      md5sum: ab3952cabf026b48cd3eb5bccbb636d1

tests/modules/bwamem2/mem/main.nf

@@ -18,9 +18,26 @@ workflow test_bwamem2_mem_single_end {
     fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
 
     BWAMEM2_INDEX ( fasta )
-    BWAMEM2_MEM ( input, BWAMEM2_INDEX.out.index )
+    BWAMEM2_MEM ( input, BWAMEM2_INDEX.out.index, false )
 }
 
+//
+// Test with single-end data and sort
+//
+workflow test_bwamem2_mem_single_end_sort {
+    input = [
+        [ id:'test', single_end:true ], // meta map
+        [
+            file(params.test_data['sarscov2']['illumina']['test_1_fastq_gz'], checkIfExists: true)
+        ]
+    ]
+    fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+
+    BWAMEM2_INDEX ( fasta )
+    BWAMEM2_MEM ( input, BWAMEM2_INDEX.out.index, true )
+}
+
+
 //
 // Test with paired-end data
 //
@@ -35,5 +52,22 @@ workflow test_bwamem2_mem_paired_end {
     fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
 
     BWAMEM2_INDEX ( fasta )
-    BWAMEM2_MEM ( input, BWAMEM2_INDEX.out.index )
+    BWAMEM2_MEM ( input, BWAMEM2_INDEX.out.index, false )
+}
+
+//
+// Test with paired-end data and sort
+//
+workflow test_bwamem2_mem_paired_end_sort {
+    input = [
+        [ id:'test', single_end:false ], // meta map
+        [
+            file(params.test_data['sarscov2']['illumina']['test_1_fastq_gz'], checkIfExists: true),
+            file(params.test_data['sarscov2']['illumina']['test_2_fastq_gz'], checkIfExists: true)
+        ]
+    ]
+    fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+
+    BWAMEM2_INDEX ( fasta )
+    BWAMEM2_MEM ( input, BWAMEM2_INDEX.out.index, true )
 }

tests/modules/bwamem2/mem/nextflow.config

@@ -2,4 +2,8 @@ process {
 
     publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
 
+    withName: BWAMEM2_MEM {
+        ext.args2 = { sort_bam ? "" : "-bh" }
+    }
+
 }

tests/modules/bwamem2/mem/test.yml

@@ -16,6 +16,24 @@
     - path: ./output/bwamem2/bwamem2/genome.fasta.ann
       md5sum: c32e11f6c859f166c7525a9c1d583567
 
+- name: bwamem2 mem single-end sort
+  command: nextflow run ./tests/modules/bwamem2/mem -entry test_bwamem2_mem_single_end_sort -c ./tests/config/nextflow.config -c ./tests/modules/bwamem2/mem/nextflow.config
+  tags:
+    - bwamem2
+    - bwamem2/mem
+  files:
+    - path: ./output/bwamem2/test.bam
+    - path: ./output/bwamem2/bwamem2/genome.fasta.amb
+      md5sum: 3a68b8b2287e07dd3f5f95f4344ba76e
+    - path: ./output/bwamem2/bwamem2/genome.fasta.pac
+      md5sum: 983e3d2cd6f36e2546e6d25a0da78d66
+    - path: ./output/bwamem2/bwamem2/genome.fasta.0123
+      md5sum: b02870de80106104abcb03cd9463e7d8
+    - path: ./output/bwamem2/bwamem2/genome.fasta.bwt.2bit.64
+      md5sum: d097a1b82dee375d41a1ea69895a9216
+    - path: ./output/bwamem2/bwamem2/genome.fasta.ann
+      md5sum: c32e11f6c859f166c7525a9c1d583567
+
 - name: bwamem2 mem paired-end
   command: nextflow run ./tests/modules/bwamem2/mem -entry test_bwamem2_mem_paired_end -c ./tests/config/nextflow.config -c ./tests/modules/bwamem2/mem/nextflow.config
   tags:
@@ -33,3 +51,21 @@
       md5sum: d097a1b82dee375d41a1ea69895a9216
     - path: ./output/bwamem2/bwamem2/genome.fasta.ann
       md5sum: c32e11f6c859f166c7525a9c1d583567
+
+- name: bwamem2 mem paired-end sort
+  command: nextflow run ./tests/modules/bwamem2/mem -entry test_bwamem2_mem_paired_end_sort -c ./tests/config/nextflow.config -c ./tests/modules/bwamem2/mem/nextflow.config
+  tags:
+    - bwamem2
+    - bwamem2/mem
+  files:
+    - path: ./output/bwamem2/test.bam
+    - path: ./output/bwamem2/bwamem2/genome.fasta.amb
+      md5sum: 3a68b8b2287e07dd3f5f95f4344ba76e
+    - path: ./output/bwamem2/bwamem2/genome.fasta.pac
+      md5sum: 983e3d2cd6f36e2546e6d25a0da78d66
+    - path: ./output/bwamem2/bwamem2/genome.fasta.0123
+      md5sum: b02870de80106104abcb03cd9463e7d8
+    - path: ./output/bwamem2/bwamem2/genome.fasta.bwt.2bit.64
+      md5sum: d097a1b82dee375d41a1ea69895a9216
+    - path: ./output/bwamem2/bwamem2/genome.fasta.ann
+      md5sum: c32e11f6c859f166c7525a9c1d583567