New fasta_bgzip_index_dict_samtools subworkflow (#11783)

* Template 4.0.2

* New `fasta_bgzip_index_dict_samtools` subworkflow

Here is a new `fasta_bgzip_index_dict_samtools` subworkflow that takes
FASTA files as input (compressed or not) and returns them
BGZF-compressed, with their flat and binary indexes, and their sequence
dictionary.

This subworkflow is especially useful when making CRAM files.

* Also collect the sequence size summary.

Author: charles-plessy

subworkflows/nf-core/fasta_bgzip_index_dict_samtools/main.nf

@@ -0,0 +1,36 @@
+include { HTSLIB_BGZIPTABIX  } from '../../../modules/nf-core/htslib/bgziptabix/main'
+include { SAMTOOLS_DICT      } from '../../../modules/nf-core/samtools/dict/main'
+include { SAMTOOLS_FAIDX     } from '../../../modules/nf-core/samtools/faidx/main'
+
+workflow FASTA_BGZIP_INDEX_DICT_SAMTOOLS {
+
+    take:
+    ch_fasta // channel: [ val(meta), fasta ]
+
+    main:
+
+    HTSLIB_BGZIPTABIX (
+        ch_fasta.map { meta, fasta -> [meta, fasta, [], []] },
+        'compress',
+        [],
+        []
+    )
+
+    SAMTOOLS_FAIDX (
+        HTSLIB_BGZIPTABIX.out.output.map {meta, fasta -> [meta, fasta, []]},
+        true
+    )
+
+    SAMTOOLS_DICT (
+        HTSLIB_BGZIPTABIX.out.output
+    )
+
+    ch_joined = HTSLIB_BGZIPTABIX.out.output
+        .join(SAMTOOLS_FAIDX.out.fai)
+        .join(SAMTOOLS_FAIDX.out.gzi)
+        .join(SAMTOOLS_FAIDX.out.sizes)
+        .join(SAMTOOLS_DICT.out.dict)
+
+    emit:
+    fasta_fai_gzi_dict = ch_joined             // channel: [ val(meta),  fasta.gz, fai, gzi, sizes, dict ]
+}

subworkflows/nf-core/fasta_bgzip_index_dict_samtools/meta.yml

@@ -0,0 +1,39 @@
+# yaml-language-server: $schema=https://raw.githubusercontent.com/nf-core/modules/master/subworkflows/yaml-schema.json
+name: "fasta_bgzip_index_dict_samtools"
+description: Ensure BGZF compression, index, and produce a sequence size summary and a sequence dictionary for each FASTA file.
+keywords:
+  - bgzip
+  - faidx
+  - dict
+  - fasta
+components:
+  - htslib/bgziptabix
+  - samtools/dict
+  - samtools/faidx
+
+input:
+  - ch_fasta:
+      type: file
+      description: |
+        FASTA file, compressed or not.
+        Structure: [ val(meta), path(fasta) ]
+      pattern: "*.{fa,fa.gz,fa.bz2,fa.xz,fasta,fasta.gz,fasta.bz2,fasta.xz}"
+
+output:
+  - fasta_fai_gzi_dict:
+      type: file
+      description: |
+        Channel containing:
+          - BGZF-compressed FASTA
+          - FASTA index (.fai)
+          - BGZF index (.gzi)
+          - sequence size summary (.sizes)
+          - sequence dictionary (.dict)
+        Structure:
+          [ val(meta), path(fasta_gz), path(fai), path(gzi), path(sizes), path(dict) ]
+      pattern: "*"
+
+authors:
+  - "@charles-plessy"
+maintainers:
+  - "@charles-plessy"

subworkflows/nf-core/fasta_bgzip_index_dict_samtools/tests/main.nf.test

@@ -0,0 +1,39 @@
+nextflow_workflow {
+
+    name "Test Subworkflow FASTA_BGZIP_INDEX_DICT_SAMTOOLS"
+    script "../main.nf"
+    workflow "FASTA_BGZIP_INDEX_DICT_SAMTOOLS"
+
+    tag "subworkflows"
+    tag "subworkflows_nfcore"
+    tag "subworkflows/fasta_bgzip_index_dict_samtools"
+    tag "htslib"
+    tag "htslib/bgziptabix"
+    tag "samtools"
+    tag "samtools/dict"
+    tag "samtools/faidx"
+
+    config "./nextflow.config"
+
+    test("sarscov2 - fasta - genomes") {
+
+        when {
+            workflow {
+                """
+                input[0] = channel.of([
+                    [ id:'genome_complete' ], // meta map
+                    file(params.modules_testdata_base_path + 'genomics/sarscov2/genome/genome.fasta', checkIfExists: true)
+                ])
+                """
+            }
+        }
+        then {
+            assert workflow.success
+            assertAll(
+                { assert snapshot(
+                    workflow.out
+                ).match() }
+            )
+        }
+    }
+}

subworkflows/nf-core/fasta_bgzip_index_dict_samtools/tests/main.nf.test.snap

@@ -0,0 +1,37 @@
+{
+    "sarscov2 - fasta - genomes": {
+        "content": [
+            {
+                "0": [
+                    [
+                        {
+                            "id": "genome_complete"
+                        },
+                        "genome_complete.gz:md5,6e9fe4042a72f2345f644f239272b7e6",
+                        "genome_complete.gz.fai:md5,9da2a56e2853dc8c0b86a9e7229c9fe5",
+                        "genome_complete.gz.gzi:md5,7dea362b3fac8e00956a4952a3d4f474",
+                        "genome_complete.gz.sizes:md5,a57c401f27ae5133823fb09fb21c8a3c",
+                        "genome_complete.gz.dict:md5,7259d9fba4f0029e294b70a7bf05af6a"
+                    ]
+                ],
+                "fasta_fai_gzi_dict": [
+                    [
+                        {
+                            "id": "genome_complete"
+                        },
+                        "genome_complete.gz:md5,6e9fe4042a72f2345f644f239272b7e6",
+                        "genome_complete.gz.fai:md5,9da2a56e2853dc8c0b86a9e7229c9fe5",
+                        "genome_complete.gz.gzi:md5,7dea362b3fac8e00956a4952a3d4f474",
+                        "genome_complete.gz.sizes:md5,a57c401f27ae5133823fb09fb21c8a3c",
+                        "genome_complete.gz.dict:md5,7259d9fba4f0029e294b70a7bf05af6a"
+                    ]
+                ]
+            }
+        ],
+        "timestamp": "2026-05-28T11:12:13.641917963",
+        "meta": {
+            "nf-test": "0.9.5",
+            "nextflow": "25.10.4"
+        }
+    }
+}

subworkflows/nf-core/fasta_bgzip_index_dict_samtools/tests/nextflow.config

@@ -0,0 +1,7 @@
+process {
+    withName: 'SAMTOOLS_DICT' {
+        // This makes the output deterministic,
+        // otherwise you get temporary file folder names in the file path.
+        ext.args = { "-u ./${fasta} -a ${meta.id}" }
+    }
+}