Fasta index bismark bwameth (#7010)

* init fasta_create_index_bismark_bwameth subworkflow * add tests + snap * Change name * update snap --------- Co-authored-by: Simon Pearce <[email protected]>
nf-core · Nov 27, 2024 · e24c973 · e24c973
1 parent 3628d82
commit e24c973
Show file tree

Hide file tree

Showing 5 changed files with 784 additions and 0 deletions.
diff --git a/subworkflows/nf-core/fasta_index_bismark_bwameth/main.nf b/subworkflows/nf-core/fasta_index_bismark_bwameth/main.nf
@@ -0,0 +1,102 @@
+include { UNTAR                     } from '../../../modules/nf-core/untar/main'
+include { GUNZIP                    } from '../../../modules/nf-core/gunzip/main'
+include { BISMARK_GENOMEPREPARATION } from '../../../modules/nf-core/bismark/genomepreparation/main'
+include { BWAMETH_INDEX             } from '../../../modules/nf-core/bwameth/index/main'
+include { SAMTOOLS_FAIDX            } from '../../../modules/nf-core/samtools/faidx/main'
+
+workflow FASTA_INDEX_BISMARK_BWAMETH {
+
+    take:
+    fasta         // channel: [ val(meta), [ fasta ] ]
+    fasta_index   // channel: [ val(meta), [ fasta index ] ]
+    bismark_index // channel: [ val(meta), [ bismark index ] ]
+    bwameth_index // channel: [ val(meta), [ bwameth index ] ]
+
+    main:
+
+    ch_fasta         = Channel.empty()
+    ch_fasta_index   = Channel.empty()
+    ch_bismark_index = Channel.empty()
+    ch_bwameth_index = Channel.empty()
+    ch_versions      = Channel.empty()
+
+    if (fasta.toString().endsWith('.gz')) {
+        GUNZIP (
+            [ [:], file(fasta, checkIfExists: true) ]
+        )
+        ch_fasta    = GUNZIP.out.gunzip
+        ch_versions = ch_versions.mix(GUNZIP.out.versions)
+    } else {
+        ch_fasta    = Channel.value([[:], file(fasta, checkIfExists: true)])
+    }
+
+    // Aligner: bismark or bismark_hisat
+    if( params.aligner =~ /bismark/ ){
+        /*
+         * Generate bismark index if not supplied
+         */
+        if (bismark_index) {
+            if (bismark_index.toString().endsWith('.gz')) {
+                UNTAR (
+                    [ [:], file(bismark_index, checkIfExists: true) ]
+                )
+                ch_bismark_index = UNTAR.out.untar
+                ch_versions      = ch_versions.mix(UNTAR.out.versions)
+            } else {
+                ch_bismark_index = Channel.value([[:], file(bismark_index, checkIfExists: true)])
+            }
+        } else {
+            BISMARK_GENOMEPREPARATION (
+                ch_fasta
+            )
+            ch_bismark_index = BISMARK_GENOMEPREPARATION.out.index
+            ch_versions      = ch_versions.mix(BISMARK_GENOMEPREPARATION.out.versions)
+        }
+    }
+
+    // Aligner: bwameth
+    else if ( params.aligner == 'bwameth' ){
+        /*
+         * Generate bwameth index if not supplied
+         */
+        if (bwameth_index) {
+            if (bwameth_index.toString().endsWith('.gz')) {
+                UNTAR (
+                    [ [:], file(bwameth_index, checkIfExists: true) ]
+                )
+                ch_bwameth_index = UNTAR.out.untar
+                ch_versions      = ch_versions.mix(UNTAR.out.versions)
+            } else {
+                ch_bwameth_index = Channel.value([[:], file(bwameth_index, checkIfExists: true)])
+            }
+        } else {
+            BWAMETH_INDEX (
+                ch_fasta
+            )
+            ch_bwameth_index = BWAMETH_INDEX.out.index
+            ch_versions      = ch_versions.mix(BWAMETH_INDEX.out.versions)
+        }
+
+        /*
+         * Generate fasta index if not supplied
+         */
+        if (fasta_index) {
+            ch_fasta_index = Channel.value(file(fasta_index, checkIfExists: true))
+        } else {
+            SAMTOOLS_FAIDX(
+                ch_fasta,
+                [[:], []]
+            )
+            ch_fasta_index = SAMTOOLS_FAIDX.out.fai
+            ch_versions    = ch_versions.mix(SAMTOOLS_FAIDX.out.versions)
+        }
+    }
+
+    emit:
+    fasta         = ch_fasta         // channel: [ val(meta), [ fasta ] ]
+    fasta_index   = ch_fasta_index   // channel: [ val(meta), [ fasta index ] ]
+    bismark_index = ch_bismark_index // channel: [ val(meta), [ bismark index ] ]
+    bwameth_index = ch_bwameth_index // channel: [ val(meta), [ bwameth index ] ]
+    versions      = ch_versions      // channel: [ versions.yml ]
+}
+
diff --git a/subworkflows/nf-core/fasta_index_bismark_bwameth/meta.yml b/subworkflows/nf-core/fasta_index_bismark_bwameth/meta.yml
@@ -0,0 +1,72 @@
+# yaml-language-server: $schema=https://raw.githubusercontent.com/nf-core/modules/master/subworkflows/yaml-schema.json
+name: "fasta_index_bismark_bwameth"
+description: Generate index files from reference fasta for bismark and bwameth
+keywords:
+  - bismark
+  - bwameth
+  - prepare genome
+  - index
+components:
+  - untar
+  - gunzip
+  - bismark/genomepreparation
+  - bwameth/index
+  - samtools/faidx
+input:
+  - fasta:
+      type: file
+      description: |
+        Reference genome
+        Structure: [ val(meta), path(fasta) ]
+      pattern: "*.{fa/fa.gz}"
+  - fasta_index:
+      type: file
+      description: |
+        Reference genome index file
+        Structure: [ val(meta), path(fasta) ]
+      pattern: "*.fai"
+  - bismark_index:
+      type: file
+      description: |
+        Bismark genome index files
+        Structure: [ val(meta), path(bismark_index) ]
+  - bwameth_index:
+      type: file
+      description: |
+        Bwameth genome index files
+        Structure: [ val(meta), path(bwameth_index) ]
+output:
+  - fasta:
+      type: file
+      description: |
+        Reference genome
+        Structure: [ val(meta), path(fasta) ]
+      pattern: "*.fa"
+  - fasta_index:
+      type: file
+      description: |
+        Reference genome index file
+        Structure: [ val(meta), path(fasta) ]
+      pattern: "*.fai"
+  - bismark_index:
+      type: file
+      description: |
+        Bismark genome index files
+        Structure: [ val(meta), path(bismark_index) ]
+        pattern: "BismarkIndex"
+  - bwameth_index:
+      type: file
+      description: |
+        Bwameth genome index files
+        Structure: [ val(meta), path(bwameth_index) ]
+        pattern: "index"
+  - versions:
+      type: file
+      description: |
+        File containing software versions
+        Structure: [ path(versions.yml) ]
+      pattern: "versions.yml"
+authors:
+  - "@sateeshperi"
+maintainers:
+  - "@sateeshperi"