feat: Add SKA2 & Gubbins option

beiko-lab · Jul 16, 2023 · 60b4664 · 60b4664
1 parent 05748a4
commit 60b4664
Show file tree

Hide file tree

Showing 7 changed files with 72 additions and 30 deletions.
diff --git a/conf/modules.config b/conf/modules.config
@@ -418,6 +418,22 @@ process {
         ]
     }
 
+    withName: SKA2 {
+        publishDir = [
+            path: { "${params.outdir}/recombination/ska2" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
+    }
+
+    withName: GUBBINS {
+        publishDir = [
+            path: { "${params.outdir}/recombination/gubbins/${cluster}" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
+    }
+
 
     withName: MULTIQC {
         publishDir = [

diff --git a/modules/local/ska.nf b/modules/local/ska.nf
@@ -1,14 +1,14 @@
-process SKA {
+process SKA2 {
     tag "$cluster"
     label 'process_medium'
 
     conda "bioconda::gubbins=3.3.0"
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/gubbins:3.3.0--py310pl5321h8472f5a_0':
-        'quay.io/biocontainers/gubbins:3.3.0--py310pl5321h8472f5a_0' }"
+        'docker://jvfe/gubbins:3.3.0':
+        'docker.io/jvfe/gubbins:3.3.0' }"
 
     input:
-    tuple val(cluster), path(assemblies), path(reference)
+    tuple val(cluster), path(assemblies), val(reference)
     path assembly_files
 
     output:

diff --git a/modules/nf-core/gubbins/main.nf b/modules/nf-core/gubbins/main.nf
@@ -7,7 +7,7 @@ process GUBBINS {
         'quay.io/biocontainers/gubbins:3.0.0--py39h5bf99c6_0' }"
 
     input:
-    path alignment
+    tuple val(cluster), path(alignment)
 
     output:
     path "*.fasta"                          , emit: fasta
@@ -29,6 +29,7 @@ process GUBBINS {
     """
     run_gubbins.py \\
         --threads $task.cpus \\
+        --prefix cluster_$cluster \\
         $args \\
         $alignment
     cat <<-END_VERSIONS > versions.yml

diff --git a/nextflow.config b/nextflow.config
@@ -48,6 +48,8 @@ params {
 
     // Recombination
     run_recombination          = false
+    run_verticall              = true
+    run_gubbins                = false
 
     // MultiQC options
     multiqc_config             = null

diff --git a/nextflow_schema.json b/nextflow_schema.json
@@ -209,6 +209,15 @@
                     "type": "boolean",
                     "description": "Run Recombination",
                     "fa_icon": "fas fa-tree"
+                },
+                "run_verticall": {
+                    "type": "boolean",
+                    "default": true,
+                    "description": "Run Verticall recombination tool"
+                },
+                "run_gubbins": {
+                    "type": "boolean",
+                    "description": "Run Gubbins recombination tool"
                 }
             }
         },

diff --git a/recombination.nf b/recombination.nf
@@ -1,6 +1,8 @@
 include { GET_RECOMB_INPUT } from './modules/local/get_recomb_input'
 include { VERTICALL_PAIRWISE } from './modules/local/verticall/pairwise.nf'
+include { SKA2 } from './modules/local/ska'
 include { QUAST } from './modules/nf-core/quast/main'
+include { GUBBINS } from './modules/nf-core/gubbins/main'
 
 workflow {
 
@@ -52,19 +54,22 @@ workflow {
             .map { row -> tuple(row.Cluster, file(row.samplesheet), row.Reference, row.reference_path) }
             .set { recomb_input }
 
-        assemblies
-            .combine (recomb_input.map { c,s,ref,reffile -> ref } | collect | map { [it] })
-            .filter { meta, path, to_remove -> !(meta["id"] in to_remove) }
-            .map { it[0, 1] }
-            .collect { meta, path -> path }
-            .set { unduplicated_assemblies }
-
-        VERTICALL_PAIRWISE (
-            recomb_input.map { c, s, r, rf -> tuple(c,s,rf) },
-            assemblies.collect { meta, path -> path }
-        )
-
-
+        if (params.run_gubbins) {
+            SKA2 (
+                recomb_input.map { c, s, r, rf -> tuple(c,s,rf) },
+                assemblies.collect { meta, path -> path }
+            )
+
+            GUBBINS (
+                SKA2.out.aln
+            )
+        }
+        if (params.run_verticall) {
+            VERTICALL_PAIRWISE (
+                recomb_input.map { c, s, r, rf -> tuple(c,s,rf) },
+                assemblies.collect { meta, path -> path }
+            )
+        }
 
     // emit:
     //     genomes_to_remove = genomes_to_remove

diff --git a/subworkflows/local/recombination.nf b/subworkflows/local/recombination.nf
@@ -1,6 +1,8 @@
 include { GET_RECOMB_INPUT } from '../../modules/local/get_recomb_input'
 include { VERTICALL_PAIRWISE } from '../../modules/local/verticall/pairwise.nf'
+include { SKA2 } from '../../modules/local/ska'
 include { QUAST } from '../../modules/nf-core/quast/main'
+include { GUBBINS } from '../../modules/nf-core/gubbins/main'
 
 workflow RECOMBINATION {
 
@@ -41,20 +43,27 @@ workflow RECOMBINATION {
             .map { row -> tuple(row.Cluster, file(row.samplesheet), row.Reference, row.reference_path) }
             .set { recomb_input }
 
-        assemblies
-            .combine (recomb_input.map { c,s,ref,reffile -> ref } | collect | map { [it] })
-            .filter { meta, path, to_remove -> !(meta["id"] in to_remove) }
-            .map { it[0, 1] }
-            .collect { meta, path -> path }
-            .set { unduplicated_assemblies }
-
-        VERTICALL_PAIRWISE (
-            recomb_input.map { c, s, r, rf -> tuple(c,s,rf) },
-            assemblies.collect { meta, path -> path }
-        )
+        if (params.run_gubbins) {
+            SKA2 (
+                recomb_input.map { c, s, r, rf -> tuple(c,s,rf) },
+                assemblies.collect { meta, path -> path }
+            )
 
+            GUBBINS (
+                SKA2.out.aln
+            )
+        }
+        if (params.run_verticall) {
+            VERTICALL_PAIRWISE (
+                recomb_input.map { c, s, r, rf -> tuple(c,s,rf) },
+                assemblies.collect { meta, path -> path }
+            )
+        }
 
+        verticall_result = VERTICALL_PAIRWISE.out.tsv.ifEmpty([])
+        gubbins_result = GUBBINS.out.stats.ifEmpty([])
 
     emit:
-        verticall_result = VERTICALL_PAIRWISE.out.tsv
+        verticall_result
+        gubbins_result
 }