Merge pull request #350 from genomic-medicine-sweden/340-add-ska-to-base

Add ska to base

CHANGELOG.md

@@ -9,6 +9,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Added
 
+- Added ska2 module processes (`ska_build` & `ska_align`)
+- Added ska2 to `bacterial_base.nf`
+
 ### Fixed
 
 - Fixed tbprofiler related installation rules in `Makefile`

configs/nextflow.base.config

@@ -235,7 +235,7 @@ process {
 	withName: samtools_index_assembly {
 		cpus = 16
 		memory = '2 GB'
-		container = "https://depot.galaxyproject.org/singularity/samtools:1.17--h00cdaf9_0"
+		container = "${params.containerDir}/samtools.sif"
 	}
 	withName: samtools_index_ref {
 		cpus = 16
@@ -254,9 +254,16 @@ process {
 	}
 	withName: shigapass {
 		cpus = 2
-		container = "https://depot.galaxyproject.org/singularity/shigapass:1.5.0--hdfd78af_0"
+		container = "${params.containerDir}/shigapass.sif"
 		publishDir = [ path: "${params.outdir}/${params.speciesDir}/shigapass", mode: 'copy', overwrite: true ]
 	}
+	withName: ska_build {
+		container = "${params.containerDir}/ska2.sif"
+		cpus = 2
+		memory = '10 GB'
+		publishDir = [ path: "${params.outdir}/${params.speciesDir}/ska", mode: 'copy', overwrite: true ]
+		ext.args = "-k 31 --threads 1"
+	}
 	withName: skesa {
 		container = "${params.containerDir}/skesa.sif"
 		cpus = 16

configs/nextflow.ci.config

@@ -207,7 +207,7 @@ process {
 	withName: samtools_index_assembly {
 		cpus = 4
 		memory = '2 GB'
-		container = "https://depot.galaxyproject.org/singularity/samtools:1.17--h00cdaf9_0"
+		container = "${params.containerDir}/samtools.sif"
 	}
 	withName: samtools_index_ref {
 		cpus = 4
@@ -226,9 +226,16 @@ process {
 	}
 	withName: shigapass {
 		cpus = 2
-		container = "https://depot.galaxyproject.org/singularity/shigapass:1.5.0--hdfd78af_0"
+		container = "${params.containerDir}/shigapass.sif"
 		publishDir = [ path: "${params.outdir}/${params.speciesDir}/shigapass", mode: 'copy', overwrite: true ]
 	}
+	withName: ska_build {
+		container = "${params.containerDir}/ska2.sif"
+		cpus = 2
+		memory = '10 GB'
+		publishDir = [ path: "${params.outdir}/${params.speciesDir}/ska", mode: 'copy', overwrite: true ]
+		ext.args = "-k 31 --threads 1"
+	}
 	withName: skesa {
 		container = "${params.containerDir}/skesa.sif"
 		cpus = 4

configs/nextflow.dev.config

@@ -244,6 +244,13 @@ process {
 		container = "https://depot.galaxyproject.org/singularity/shigapass:1.5.0--hdfd78af_0"
 		publishDir = [ path: "${params.outdir}/${params.speciesDir}/shigapass", mode: 'copy', overwrite: true ]
 	}
+	withName: ska_build {
+		container = "https://depot.galaxyproject.org/singularity/ska2:0.3.10--h4349ce8_0"
+		cpus = 2
+		memory = '10 GB'
+		publishDir = [ path: "${params.outdir}/${params.speciesDir}/ska", mode: 'copy', overwrite: true ]
+		ext.args = "-k 31 --threads 1"
+	}
 	withName: skesa {
 		container = "https://depot.galaxyproject.org/singularity/skesa:2.4.0--he1c1bb9_0"
 		cpus = 16

configs/nextflow.hopper.config

@@ -259,6 +259,13 @@ process {
 		container = "https://depot.galaxyproject.org/singularity/shigapass:1.5.0--hdfd78af_0"
 		publishDir = [ path: "${params.outdir}/${params.speciesDir}/shigapass", mode: 'copy', overwrite: true ]
 	}
+	withName: ska_build {
+		container = "https://depot.galaxyproject.org/singularity/ska2:0.3.10--h4349ce8_0"
+		cpus = 2
+		memory = '10 GB'
+		publishDir = [ path: "${params.outdir}/${params.speciesDir}/ska", mode: 'copy', overwrite: true ]
+		ext.args = "-k 31 --threads 1"
+	}
 	withName: skesa {
 		container = "https://depot.galaxyproject.org/singularity/skesa:2.4.0--he1c1bb9_0"
 		cpus = 16

configs/nextflow.ngp.config

@@ -235,7 +235,7 @@ process {
 	withName: samtools_index_assembly {
 		cpus = params.cpus_large
 		memory = params.mem_large
-		container = "https://depot.galaxyproject.org/singularity/samtools:1.17--h00cdaf9_0"
+		container = "${params.containerDir}/samtools.sif"
 	}
 	withName: samtools_index_ref {
 		cpus = params.cpus_large
@@ -246,7 +246,7 @@ process {
 	withName: samtools_sort_ref {
 		cpus = params.cpus_large
 		memory = params.mem_small
-		container = "https://depot.galaxyproject.org/singularity/samtools:1.17--h00cdaf9_0"
+		container = "${params.containerDir}/samtools.sif"
 		publishDir = [ path: "${params.outdir}/${params.speciesDir}/${params.bamDir}", mode: 'copy', overwrite: true ]
 	}
 	withName: save_analysis_metadata {
@@ -259,10 +259,17 @@ process {
 		publishDir = [ path: "${params.outdir}/${params.speciesDir}/serotypefinder", mode: 'copy', overwrite: true ]
 	}
 	withName: shigapass {
-		cpus = 2
-		container = "https://depot.galaxyproject.org/singularity/shigapass:1.5.0--hdfd78af_0"
+		cpus = params.cpus_small
+		container = "${params.containerDir}/shigapass.sif"
 		publishDir = [ path: "${params.outdir}/${params.speciesDir}/shigapass", mode: 'copy', overwrite: true ]
 	}
+	withName: ska_build {
+		container = "${params.containerDir}/ska2.sif"
+		cpus = params.cpus_small
+		memory = params.mem_small
+		publishDir = [ path: "${params.outdir}/${params.speciesDir}/ska", mode: 'copy', overwrite: true ]
+		ext.args = "-k 31 --threads 1"
+	}
 	withName: skesa {
 		container = "${params.containerDir}/skesa.sif"
 		cpus = params.cpus_large

container/Makefile

@@ -34,6 +34,7 @@ remote_containers := ncbi-amrfinderplus.sif \
 					samtools.sif \
 					serotypefinder.sif \
 					shigapass.sif \
+					ska2.sif \
 					skesa.sif \
 					snippy.sif \
 					sourmash.sif \
@@ -62,6 +63,7 @@ URL_resfinder := https://depot.galaxyproject.org/singularity/resfinder:4.4.2--py
 URL_samtools := https://depot.galaxyproject.org/singularity/samtools:1.17--h00cdaf9_0
 URL_serotypefinder := https://depot.galaxyproject.org/singularity/serotypefinder:2.0.2--py312hdfd78af_1
 URL_shigapass := https://depot.galaxyproject.org/singularity/shigapass:1.5.0--hdfd78af_0
+URL_ska2 := https://depot.galaxyproject.org/singularity/ska2:0.3.10--h4349ce8_0
 URL_skesa := https://depot.galaxyproject.org/singularity/skesa:2.5.1--hdcf5f25_0
 URL_snippy := https://depot.galaxyproject.org/singularity/snippy:4.6.0--hdfd78af_2
 URL_sourmash := https://depot.galaxyproject.org/singularity/sourmash:4.8.2--hdfd78af_0

nextflow-modules/modules/ska/main.nf

@@ -0,0 +1,42 @@
+process ska_build {
+  tag "${sampleID}"
+  scratch params.scratch
+
+  input:
+    tuple val(sampleID), path(reads)
+
+  output:
+    tuple val(sampleID), path(output), emit: skf
+    path "*versions.yml"             , emit: versions
+
+  script:
+    def args = task.ext.args ?: ''
+    def inputData = reads.size() == 2 ? "${sampleID}\t${reads[0]}\t${reads[1]}" : "${sampleID}\t${reads[0]}"
+    output_basename = "${sampleID}_ska_index"
+    output = "${output_basename}.skf"
+    """
+    echo ${inputData} > ${sampleID}_input.txt
+    ska build $args -o ${output_basename} -f ${sampleID}_input.txt
+
+    cat <<-END_VERSIONS > ${sampleID}_${task.process}_versions.yml
+    ${task.process}:
+     ska2:
+      version: \$(echo \$(ska --version 2>&1) | sed 's/^.*ska // ; s/ .*//')
+      container: ${task.container}
+    END_VERSIONS
+    """
+
+  stub:
+    output = "${sampleID}_ska_index.skf"
+    """
+    mkdir ${sampleID}
+    touch $output
+
+    cat <<-END_VERSIONS > ${sampleID}_${task.process}_versions.yml
+    ${task.process}:
+     ska2:
+      version: \$(echo \$(ska --version 2>&1) | sed 's/^.*ska // ; s/ .*//')
+      container: ${task.container}
+    END_VERSIONS
+    """
+}

workflows/bacterial_base.nf

@@ -11,6 +11,7 @@ include { post_align_qc                         } from '../nextflow-modules/modu
 include { quast                                 } from '../nextflow-modules/modules/quast/main.nf'
 include { samtools_index as samtools_index_ref  } from '../nextflow-modules/modules/samtools/main.nf'
 include { save_analysis_metadata                } from '../nextflow-modules/modules/meta/main.nf'
+include { ska_build                             } from '../nextflow-modules/modules/ska/main.nf'
 include { skesa                                 } from '../nextflow-modules/modules/skesa/main.nf'
 include { sourmash                              } from '../nextflow-modules/modules/sourmash/main.nf'
 include { spades_illumina                       } from '../nextflow-modules/modules/spades/main.nf'
@@ -64,12 +65,15 @@ workflow CALL_BACTERIAL_BASE {
 
         sourmash(ch_assembly)
 
+        ska_build(ch_reads)
+
         ch_versions = ch_versions.mix(bwa_mem_ref.out.versions)
         ch_versions = ch_versions.mix(flye.out.versions)
         ch_versions = ch_versions.mix(medaka.out.versions)
         ch_versions = ch_versions.mix(quast.out.versions)
         ch_versions = ch_versions.mix(samtools_index_ref.out.versions)
         ch_versions = ch_versions.mix(skesa.out.versions)
+        ch_versions = ch_versions.mix(ska_build.out.versions)
         ch_versions = ch_versions.mix(sourmash.out.versions)
         ch_versions = ch_versions.mix(spades_illumina.out.versions)
         ch_versions = ch_versions.mix(spades_iontorrent.out.versions)
@@ -83,6 +87,7 @@ workflow CALL_BACTERIAL_BASE {
         qc          = post_align_qc.out.qc              // channel: [ val(meta), path(fasta)]
         quast       = quast.out.qc                      // channel: [ val(meta), path(qc)]
         reads       = ch_reads                          // channel: [ val(meta), path(json)]
+        ska_build   = ska_build.out.skf                 // channel: [ val(meta), path(skf)]
         seqrun_meta = ch_seqrun_meta                    // channel: [ val(meta), val(json), val(json)]
         sourmash    = sourmash.out.signature            // channel: [ val(meta), path(signature)]
         versions    = ch_versions                       // channel: [ versions.yml ]