Implement hostremoval

nextflow.config

@@ -81,12 +81,18 @@ params {
     save_runmerged_reads        = false
 
     // Host Removal
-    perform_shortread_hostremoval     = false
+    perform_shortread_hostremoval      = false
-    shortread_hostremoval_reference   = null
+    shortread_hostremoval_reference    = null
-    shortread_hostremoval_index       = null
+    shortread_hostremoval_index        = null
-    save_hostremoval_index            = false
+    longread_hostremoval_index         = null
-    save_hostremoval_mapped           = false
+    save_hostremoval_index             = false
-    save_hostremoval_unmapped         = false
+    save_hostremoval_mapped            = false
+    save_hostremoval_unmapped          = false
+    save_minimap2_hostremoval_index    = false
+    save_minimap2_hostremoval_mapped   = false
+    save_minimap2_hostremoval_unmapped = false
+    save_samtools_unmapped_bam         = false
+    save_minimap2_unmapped_fq          = false
 
     // MALT
     run_malt                   = false

subworkflows/local/longread_hostremoval.nf

@@ -0,0 +1,47 @@
+//
+// Remove host reads via alignment and export off-target reads
+//
+
+include { MINIMAP2_INDEX             } from '../../modules/nf-core/modules/minimap2/index/main'
+include { MINIMAP2_ALIGN             } from '../../modules/nf-core/modules/minimap2/align/main'
+include { SAMTOOLS_VIEW              } from '../../modules/nf-core/modules/samtools/view/main'
+include { SAMTOOLS_BAM2FQ            } from '../../modules/nf-core/modules/samtools/bam2fq/main'
+
+workflow LONGREAD_HOSTREMOVAL {
+    take:
+    reads     // [ [ meta ], [ reads ] ]
+    reference // /path/to/fasta
+    index     // /path/to/index
+
+    main:
+    ch_versions       = Channel.empty()
+    ch_multiqc_files  = Channel.empty()
+
+    if ( !params.longread_hostremoval_index ) {
+        ch_minimap2_index = MINIMAP2_INDEX ( reference ).index
+        ch_versions       = ch_versions.mix( MINIMAP2_INDEX.out.versions )
+    } else {
+        ch_minimap2_index = index
+    }
+
+    MINIMAP2_ALIGN ( reads, ch_minimap2_index, true, false, false )
+    ch_versions        = ch_versions.mix( MINIMAP2_ALIGN.out.versions.first() )
+    ch_minimap2_mapped = MINIMAP2_ALIGN.out.bam
+        .map {
+            meta, reads ->
+                [ meta, reads, [] ]
+        }
+
+
+    SAMTOOLS_VIEW ( ch_minimap2_mapped , [] )
+    ch_versions      = ch_versions.mix( SAMTOOLS_VIEW.out.versions.first() )
+
+    SAMTOOLS_BAM2FQ ( SAMTOOLS_VIEW.out.bam, false )
+    ch_versions      = ch_versions.mix( SAMTOOLS_BAM2FQ.out.versions.first() )
+
+
+    emit:
+    reads    = SAMTOOLS_BAM2FQ.out.reads   // channel: [ val(meta), [ reads ] ]
+    versions = ch_versions                 // channel: [ versions.yml ]
+}
+

workflows/taxprofiler.nf

@@ -26,7 +26,8 @@ if (params.perform_shortread_hostremoval && !params.shortread_hostremoval_refere
 if (!params.shortread_hostremoval_reference && params.shortread_hostremoval_reference_index) { exit 1, "ERROR: [nf-core/taxprofiler] --shortread_hostremoval_index provided but no --shortread_hostremoval_reference FASTA supplied. Check input." }
 
 if (params.shortread_hostremoval_reference ) { ch_reference       = file(params.shortread_hostremoval_reference) }
-if (params.shortread_hostremoval_index     ) { ch_reference_index = file(params.shortread_hostremoval_index    ) } else { ch_reference_index = [] }
+if (params.shortread_hostremoval_index     ) { ch_shortread_reference_index = file(params.shortread_hostremoval_index    ) } else { ch_shortread_reference_index = [] }
+if (params.longread_hostremoval_index      ) { ch_longread_reference_index  = file(params.longread_hostremoval_index     ) } else { ch_longread_reference_index  = [] }
 
 /*
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
@@ -52,6 +53,7 @@ include { DB_CHECK                      } from '../subworkflows/local/db_check'
 include { SHORTREAD_PREPROCESSING       } from '../subworkflows/local/shortread_preprocessing'
 include { LONGREAD_PREPROCESSING        } from '../subworkflows/local/longread_preprocessing'
 include { SHORTREAD_HOSTREMOVAL         } from '../subworkflows/local/shortread_hostremoval'
+include { LONGREAD_HOSTREMOVAL          } from '../subworkflows/local/longread_hostremoval'
 include { SHORTREAD_COMPLEXITYFILTERING } from '../subworkflows/local/shortread_complexityfiltering'
 include { PROFILING                     } from '../subworkflows/local/profiling'
 
@@ -141,16 +143,23 @@ workflow TAXPROFILER {
     */
 
     if ( params.perform_shortread_hostremoval ) {
-        ch_shortreads_hostremoved = SHORTREAD_HOSTREMOVAL ( ch_shortreads_filtered, ch_reference, ch_reference_index ).reads
+        ch_shortreads_hostremoved = SHORTREAD_HOSTREMOVAL ( ch_shortreads_filtered, ch_reference, ch_shortread_reference_index ).reads
         ch_versions = ch_versions.mix(SHORTREAD_HOSTREMOVAL.out.versions)
     } else {
         ch_shortreads_hostremoved = ch_shortreads_filtered
     }
 
+    if ( params.perform_longread_hostremoval ) {
+        ch_longreads_hostremoved = LONGREAD_HOSTREMOVAL ( ch_longreads_preprocessed, ch_reference, ch_longread_reference_index ).reads
+        ch_versions = ch_versions.mix(LONGREAD_HOSTREMOVAL.out.versions)
+    } else {
+        ch_longreads_hostremoved = ch_longreads_preprocessed
+    }
+
     if ( params.perform_runmerging ) {
 
         ch_reads_for_cat_branch = ch_shortreads_hostremoved
-            .mix( ch_longreads_preprocessed )
+            .mix( ch_longreads_hostremoved )
             .map {
                 meta, reads ->
                     def meta_new = meta.clone()
@@ -182,7 +191,7 @@ workflow TAXPROFILER {
 
     } else {
         ch_reads_runmerged = ch_shortreads_hostremoved
-            .mix( ch_longreads_preprocessed, INPUT_CHECK.out.fasta )
+            .mix( ch_longreads_hostremoved, INPUT_CHECK.out.fasta )
     }
 
     /*