Prepare genome subworkflow

conf/modules.config

@@ -20,6 +20,14 @@ process {
         ]
     }
 
+    withName: "WINDOWMASKER_MKCOUNTS" {
+        ext.args = "-infmt fasta -sformat obinary"
+    }
+
+    withName: "WINDOWMASKER_USTAT" {
+        ext.args = "-infmt fasta -dust T -outfmt fasta"
+    }
+
     withName: "MINIMAP2_HIC" {
         ext.args = "-ax sr"
     }
@@ -40,6 +48,10 @@ process {
         ext.args = "-ax map-ont"
     }
 
+    withName: "SAMTOOLS_VIEW" {
+        ext.args = "--output-fmt bam --write-index"
+    }
+
     withName: "SAMTOOLS_INDEX" {
         ext.args = "-c"
     }
@@ -48,10 +60,6 @@ process {
         ext.args = "--lineage --busco"
     }
 
-    withName: "SAMTOOLS_VIEW" {
-        ext.args = "--output-fmt bam --write-index"
-    }
-
     withName: "BUSCO" {
         scratch = true
         // Overridden in the test profile, see at the end of this file

modules.json

@@ -65,6 +65,16 @@
                         "branch": "master",
                         "git_sha": "3ffae3598260a99e8db3207dead9f73f87f90d1f",
                         "installed_by": ["modules"]
+                    },
+                    "windowmasker/mkcounts": {
+                        "branch": "master",
+                        "git_sha": "30c3ed32e8bd5ddaf349ba2f4f99d38182fdc08c",
+                        "installed_by": ["modules"]
+                    },
+                    "windowmasker/ustat": {
+                        "branch": "master",
+                        "git_sha": "726ee59cd9360a965d96ea9ea8770f16b8ddd6cc",
+                        "installed_by": ["modules"]
                     }
                 }
             },

nextflow.config

@@ -14,6 +14,7 @@ params {
     input                      = null
     yaml                       = null
     align                      = false
+    mask                       = false
 
     // Reference options    
     fasta                      = null

nextflow_schema.json

@@ -24,7 +24,13 @@
                 },
                 "align": {
                     "type": "boolean",
-                    "description": "Boolean to turn on optional alignment before running the rest of the pipeline."
+                    "description": "Turn on optional alignment before running the rest of the pipeline.",
+                    "fa_icon": "fas fa-toggle-off"
+                },
+                "mask": {
+                    "type": "boolean",
+                    "description": "Turn on optional genome masking if needed.",
+                    "fa_icon": "fas fa-toggle-off"
                 },
                 "yaml": {
                     "type": "string",

subworkflows/local/prepare_genome.nf

@@ -0,0 +1,49 @@
+//
+// Prepare genome for downstream processing
+//
+
+include { GUNZIP                } from '../../modules/nf-core/gunzip/main'
+include { WINDOWMASKER_MKCOUNTS } from '../../modules/nf-core/windowmasker/mkcounts/main'
+include { WINDOWMASKER_USTAT    } from '../../modules/nf-core/windowmasker/ustat/main'
+
+
+workflow PREPARE_GENOME {
+    take:
+    fasta     // channel: [ meta, path(genome) ]
+
+
+    main:
+    ch_versions = Channel.empty()
+
+
+    //
+    // MODULE: Decompress FASTA file if needed
+    //
+    if ( params.fasta.endsWith('.gz') ) {
+        ch_genome   = GUNZIP ( fasta ).gunzip
+        ch_versions = ch_versions.mix ( GUNZIP.out.versions )
+    } else {
+        ch_genome   = fasta
+    }
+
+
+    //
+    // MODULES: Mask the genome if needed
+    //
+    if ( params.mask ) {
+        WINDOWMASKER_MKCOUNTS ( ch_genome )
+        ch_versions = ch_versions.mix ( WINDOWMASKER_MKCOUNTS.out.versions )
+
+        WINDOWMASKER_USTAT ( WINDOWMASKER_MKCOUNTS.out.counts, ch_genome )
+        ch_versions = ch_versions.mix ( WINDOWMASKER_USTAT.out.versions )
+
+        ch_fasta = WINDOWMASKER_USTAT.out.intervals
+    } else {
+        ch_fasta = ch_genome
+    }
+
+
+    emit:
+    genome   = ch_fasta            // channel: [ meta, path(genome) ]
+    versions = ch_versions         // channel: [ versions.yml ]
+}