feat(fastq_qc_trim_filter_setstrandedness): Add Bowtie2 as alternative rRNA removal tool

Add bowtie2 as a third option for rRNA removal alongside sortmerna and ribodetector.

Implementation details:
- Paired-end: Uses samtools view -f 12 to filter pairs where BOTH mates are unmapped
  (bowtie2's --un-conc-gz incorrectly includes pairs where one mate aligned)
- Single-end: Uses bowtie2's --un-gz directly via save_unaligned=true
- Converts U→T in rRNA reference FASTAs (RNA sequences contain U, reads contain T)

Changes:
- Add BOWTIE2_ALIGN, BOWTIE2_ALIGN_PE, BOWTIE2_BUILD module imports
- Add SAMTOOLS_VIEW and SAMTOOLS_FASTQ for paired-end filtering
- Add ch_bowtie2_index input and make_bowtie2_index parameter
- Update meta.yml with bowtie2 in ribo_removal_tool enum
- Add paired-end and single-end bowtie2 test cases

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <[email protected]>

Author: pinin4fjords

subworkflows/nf-core/fastq_qc_trim_filter_setstrandedness/main.nf

@@ -1,5 +1,10 @@
 include { BBMAP_BBSPLIT                         } from '../../../modules/nf-core/bbmap/bbsplit'
+include { BOWTIE2_ALIGN                         } from '../../../modules/nf-core/bowtie2/align/main'
+include { BOWTIE2_ALIGN as BOWTIE2_ALIGN_PE     } from '../../../modules/nf-core/bowtie2/align/main'
+include { BOWTIE2_BUILD                         } from '../../../modules/nf-core/bowtie2/build/main'
 include { CAT_FASTQ                             } from '../../../modules/nf-core/cat/fastq/main'
+include { SAMTOOLS_VIEW as SAMTOOLS_VIEW_BOWTIE2 } from '../../../modules/nf-core/samtools/view/main'
+include { SAMTOOLS_FASTQ as SAMTOOLS_FASTQ_BOWTIE2 } from '../../../modules/nf-core/samtools/fastq/main'
 include { RIBODETECTOR                          } from '../../../modules/nf-core/ribodetector/main'
 include { SEQKIT_STATS                          } from '../../../modules/nf-core/seqkit/stats/main'
 include { SORTMERNA                             } from '../../../modules/nf-core/sortmerna/main'
@@ -116,8 +121,9 @@ workflow FASTQ_QC_TRIM_FILTER_SETSTRANDEDNESS {
     ch_gtf               // channel: /path/to/genome.gtf
     ch_salmon_index      // channel: /path/to/salmon/index/ (optional)
     ch_sortmerna_index   // channel: /path/to/sortmerna/index/ (optional)
+    ch_bowtie2_index     // channel: /path/to/bowtie2/index/ (optional)
     ch_bbsplit_index     // channel: /path/to/bbsplit/index/ (optional)
-    ch_rrna_fastas       // channel: one or more fasta files containing rrna sequences to be passed to SortMeRNA (optional)
+    ch_rrna_fastas       // channel: one or more fasta files containing rrna sequences to be passed to SortMeRNA/Bowtie2 (optional)
 
     // Skip options
     skip_bbsplit         // boolean: Skip BBSplit for removal of non-reference genome reads.
@@ -129,6 +135,7 @@ workflow FASTQ_QC_TRIM_FILTER_SETSTRANDEDNESS {
     // Index generation
     make_salmon_index    // boolean: Whether to create salmon index before running salmon quant
     make_sortmerna_index // boolean: Whether to create a sortmerna index before running sortmerna
+    make_bowtie2_index   // boolean: Whether to create a bowtie2 index before running bowtie2
 
     // Trimming options
     trimmer              // string (enum): 'fastp' or 'trimgalore'
@@ -138,7 +145,7 @@ workflow FASTQ_QC_TRIM_FILTER_SETSTRANDEDNESS {
 
     // rRNA removal options
     remove_ribo_rna           // boolean: true/false: whether to remove rRNA
-    ribo_removal_tool         // string (enum): 'sortmerna' or 'ribodetector'
+    ribo_removal_tool         // string (enum): 'sortmerna', 'ribodetector', or 'bowtie2'
 
     // UMI options
     with_umi             // boolean: true/false: Enable UMI-based read deduplication.
@@ -352,6 +359,86 @@ workflow FASTQ_QC_TRIM_FILTER_SETSTRANDEDNESS {
 
             ch_versions = ch_versions.mix(RIBODETECTOR.out.versions.first())
         }
+        else if (ribo_removal_tool == 'bowtie2') {
+            if (make_bowtie2_index) {
+                // Collect all fastas into a single file for index building
+                // Convert U to T since rRNA references may contain RNA (U) but reads are DNA (T)
+                ch_rrna_fastas
+                    .collectFile(name: 'rrna_combined.fasta', newLine: true)
+                    .map { fasta ->
+                        def content = fasta.text.replaceAll('U', 'T').replaceAll('u', 't')
+                        def convertedFasta = file("${fasta.parent}/rrna_combined_dna.fasta")
+                        convertedFasta.text = content
+                        [[id: 'rrna_refs'], convertedFasta]
+                    }
+                    .set { ch_combined_fasta }
+
+                BOWTIE2_BUILD(
+                    ch_combined_fasta
+                )
+                ch_bowtie2_index = BOWTIE2_BUILD.out.index.first()
+                ch_versions = ch_versions.mix(BOWTIE2_BUILD.out.versions.first())
+            }
+
+            // Branch reads by single-end vs paired-end for different filtering strategies
+            ch_filtered_reads
+                .branch { meta, reads ->
+                    single_end: meta.single_end
+                    paired_end: !meta.single_end
+                }
+                .set { ch_reads_for_bowtie2 }
+
+            // For single-end reads: bowtie2's --un-gz works correctly
+            // save_unaligned=true outputs unmapped reads directly
+            BOWTIE2_ALIGN(
+                ch_reads_for_bowtie2.single_end,
+                ch_bowtie2_index,
+                [[], []],  // No reference fasta needed
+                true,      // save_unaligned - for single-end this works correctly
+                false,     // sort_bam - not needed
+            )
+
+            ch_multiqc_files = ch_multiqc_files.mix(BOWTIE2_ALIGN.out.log)
+            ch_versions = ch_versions.mix(BOWTIE2_ALIGN.out.versions.first())
+
+            // For paired-end reads: bowtie2's --un-conc-gz outputs pairs that didn't
+            // align concordantly, which INCLUDES pairs where one mate aligned.
+            // We need to filter via samtools to get pairs where BOTH mates are unmapped.
+            BOWTIE2_ALIGN_PE(
+                ch_reads_for_bowtie2.paired_end,
+                ch_bowtie2_index,
+                [[], []],  // No reference fasta needed for BAM output
+                false,     // save_unaligned - we'll extract from BAM instead
+                false,     // sort_bam - not needed
+            )
+
+            ch_multiqc_files = ch_multiqc_files.mix(BOWTIE2_ALIGN_PE.out.log)
+            ch_versions = ch_versions.mix(BOWTIE2_ALIGN_PE.out.versions.first())
+
+            // Filter BAM for read pairs where BOTH mates are unmapped (flag 12 = 4 + 8)
+            // This removes any pair where at least one mate aligned to rRNA
+            SAMTOOLS_VIEW_BOWTIE2(
+                BOWTIE2_ALIGN_PE.out.bam.map { meta, bam -> [meta, bam, []] },
+                [[], []],  // No reference fasta
+                [],        // No qname file
+                []         // No index format
+            )
+
+            ch_versions = ch_versions.mix(SAMTOOLS_VIEW_BOWTIE2.out.versions.first())
+
+            // Convert filtered BAM back to paired FASTQ
+            SAMTOOLS_FASTQ_BOWTIE2(
+                SAMTOOLS_VIEW_BOWTIE2.out.bam,
+                false  // not interleaved
+            )
+
+            ch_versions = ch_versions.mix(SAMTOOLS_FASTQ_BOWTIE2.out.versions.first())
+
+            // Combine single-end and paired-end results
+            BOWTIE2_ALIGN.out.fastq
+                .mix(SAMTOOLS_FASTQ_BOWTIE2.out.fastq)
+                .set { ch_filtered_reads }
+        }
 
         if (!skip_linting) {
             FQ_LINT_AFTER_RIBO_REMOVAL(

subworkflows/nf-core/fastq_qc_trim_filter_setstrandedness/meta.yml

@@ -9,8 +9,12 @@ keywords:
   - strandedness
 components:
   - bbmap/bbsplit
+  - bowtie2/align
+  - bowtie2/build
+  - samtools/fastq
   - samtools/sort
   - samtools/index
+  - samtools/view
   - cat
   - cat/fastq
   - fq/lint
@@ -79,6 +83,15 @@ input:
         - index:
             type: directory
             description: SortMeRNA index directory
+  - ch_bowtie2_index:
+      description: Directory containing bowtie2 index for rRNA removal
+      structure:
+        - meta:
+            type: map
+            description: Metadata for the Bowtie2 index
+        - index:
+            type: directory
+            description: Bowtie2 index directory
   - ch_bbsplit_index:
       description: Path to directory or tar.gz archive for pre-built BBSplit index
       structure:
@@ -90,7 +103,7 @@ input:
             description: BBSplit index directory or tar.gz archive
             pattern: "{*,*.tar.gz}"
   - ch_rrna_fastas:
-      description: Channel containing one or more FASTA files containing rRNA sequences for use with SortMeRNA
+      description: Channel containing one or more FASTA files containing rRNA sequences for use with SortMeRNA or Bowtie2
       structure:
         - meta:
             type: map
@@ -120,6 +133,9 @@ input:
   - make_sortmerna_index:
       type: boolean
       description: Whether to create sortmerna index before running sortmerna
+  - make_bowtie2_index:
+      type: boolean
+      description: Whether to create bowtie2 index before running bowtie2 for rRNA removal
   - trimmer:
       type: string
       description: Specifies the trimming tool to use
@@ -140,7 +156,7 @@ input:
   - ribo_removal_tool:
       type: string
       description: Specifies the rRNA removal tool to use
-      enum: ["sortmerna", "ribodetector"]
+      enum: ["sortmerna", "ribodetector", "bowtie2"]
   - with_umi:
       type: boolean
       description: Enable UMI-based read deduplication