Align parabricks subworkflow

subworkflows/nf-core/fastq_align_parabricks/main.nf

@@ -0,0 +1,48 @@
+//
+// Alignment and BQSR with Nvidia CLARA Parabricks
+//
+
+include { PARABRICKS_FQ2BAM    } from '../../../modules/nf-core/parabricks/fq2bam/main'
+include { PARABRICKS_APPLYBQSR } from '../../../modules/nf-core/parabricks/applybqsr/main'
+
+workflow FASTQ_ALIGN_PARABRICKS {
+
+    take:
+    ch_reads // channel: [mandatory] meta, reads
+    ch_fasta // channel: [mandatory] meta, fasta
+    ch_index // channel: [mandatory] meta, index
+    ch_interval_file // channel: [optional for parabricks] meta, intervals_bed_combined
+    ch_known_sites // channel [optional for parabricks] known_sites_indels
+
+    main:
+    ch_versions = Channel.empty()
+    ch_bam = Channel.empty()
+    ch_bai = Channel.empty()
+    ch_bqsr_table = Channel.empty()
+    ch_qc_metrics = Channel.empty()
+    ch_duplicate_metrics = Channel.empty()
+
+    PARABRICKS_FQ2BAM(ch_reads, ch_fasta, ch_index, ch_interval_file, ch_known_sites)
+
+    // Collecting FQ2BAM outputs
+    ch_qc_metrics = ch_qc_metrics.mix(PARABRICKS_FQ2BAM.out.qc_metrics)
+    ch_duplicate_metrics = ch_duplicate_metrics.mix(PARABRICKS_FQ2BAM.out.duplicate_metrics)
+
+    // Apply BQSR
+    PARABRICKS_APPLYBQSR(
+        PARABRICKS_FQ2BAM.out.bam, 
+        PARABRICKS_FQ2BAM.out.bai,
+        PARABRICKS_FQ2BAM.out.bqsr_table,
+        ch_interval_file, 
+        ch_fasta
+        )
+
+    ch_versions = ch_versions.mix(PARABRICKS_FQ2BAM.out.versions)
+    ch_versions = ch_versions.mix(PARABRICKS_APPLYBQSR.out.versions)
+
+    emit:
+    bam = PARABRICKS_APPLYBQSR.out.bam      // channel: [ [meta], bam ]
+    bai = PARABRICKS_APPLYBQSR.out.bai      // channel: [ [meta], bai ]
+    versions = ch_versions                  // channel: [ versions.yml ]
+
+}

subworkflows/nf-core/fastq_align_parabricks/meta.yml

@@ -0,0 +1,51 @@
+# yaml-language-server: $schema=https://raw.githubusercontent.com/nf-core/modules/master/subworkflows/yaml-schema.json
+name: "fastq_align_parabricks"
+## TODO nf-core: Add a description of the subworkflow and list keywords
+description: Sort SAM/BAM/CRAM file
+keywords:
+  - sort
+  - bam
+  - sam
+  - cram
+## TODO nf-core: Add a list of the modules and/or subworkflows used in the subworkflow
+components:
+  - parabricks/fq2bam
+  - parabricks/applybqsr
+## TODO nf-core: List all of the channels used as input with a description and their structure
+input:
+  - ch_bam:
+      type: file
+      description: |
+        The input channel containing the BAM/CRAM/SAM files
+        Structure: [ val(meta), path(bam) ]
+      pattern: "*.{bam/cram/sam}"
+## TODO nf-core: List all of the channels used as output with a descriptions and their structure
+output:
+  - bam:
+      type: file
+      description: |
+        Channel containing BAM files
+        Structure: [ val(meta), path(bam) ]
+      pattern: "*.bam"
+  - bai:
+      type: file
+      description: |
+        Channel containing indexed BAM (BAI) files
+        Structure: [ val(meta), path(bai) ]
+      pattern: "*.bai"
+  - csi:
+      type: file
+      description: |
+        Channel containing CSI files
+        Structure: [ val(meta), path(csi) ]
+      pattern: "*.csi"
+  - versions:
+      type: file
+      description: |
+        File containing software versions
+        Structure: [ path(versions.yml) ]
+      pattern: "versions.yml"
+authors:
+  - "@famosab"
+maintainers:
+  - "@famosab"

subworkflows/nf-core/fastq_align_parabricks/tests/lowmem.config

@@ -0,0 +1,13 @@
+process {
+
+    withName: 'PARABRICKS_FQ2BAM' {
+        ext.args = '--low-memory'
+    }
+
+    withName: 'PARABRICKS_APPLYBQSR' {
+        stageInMode = 'copy'
+    }
+    // Ref: https://forums.developer.nvidia.com/t/problem-with-gpu/256825/6
+    // Parabricks’s fq2bam requires 24GB of memory.
+    // Using --low-memory for testing
+}

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

@@ -0,0 +1,105 @@
+nextflow_workflow {
+
+    name "Test Subworkflow FASTQ_ALIGN_PARABRICKS"
+    script "../main.nf"
+    workflow "FASTQ_ALIGN_PARABRICKS"
+
+    tag "subworkflows"
+    tag "subworkflows_nfcore"
+    tag "subworkflows/fastq_align_parabricks"
+    tag "parabricks"
+    tag "parabricks/fq2bam"
+    tag "parabricks/applybqsr"
+    tag "bwa"
+    tag "bwa/index"
+
+    config "./lowmem.config"
+
+    setup {
+        run("BWA_INDEX") {
+            script "../../../../modules/nf-core/bwa/index/main.nf"
+            process {
+                """
+                input[0] = Channel.of([
+                    [ id:'test' ], // meta map
+                    file(params.modules_testdata_base_path + 'genomics/sarscov2/genome/genome.fasta', checkIfExists: true)
+                ])
+                """
+            }
+        }
+    }
+
+    test("sarscov2 - fastq.gz - single end") {
+
+        when {
+            workflow {
+                """
+                input[0] = Channel.of([
+                        [ id:'test', single_end:true ],
+                        [ file(params.modules_testdata_base_path + 'genomics/sarscov2/illumina/fastq/test_1.fastq.gz', checkIfExists: true)]
+                        ])
+                input[1] = Channel.value([
+                        [id: 'reference'],
+                        file(params.modules_testdata_base_path + 'genomics/sarscov2/genome/genome.fasta', checkIfExists: true)
+                        ])
+                input[2] = BWA_INDEX.out.index
+                input[3] = Channel.value([
+                        [id: 'intervals'],
+                        file(params.modules_testdata_base_path + 'genomics/sarscov2/genome/picard/baits.interval_list', checkIfExists: true)
+                        ])
+                input[4] = file(params.modules_testdata_base_path + 'genomics/sarscov2/illumina/vcf/test.vcf.gz', checkIfExists: true)
+                """
+            }
+        }
+
+        then {
+            assertAll(
+                { assert workflow.success},
+                { assert snapshot(
+                    bam(workflow.out.bam[0][1]).getReadsMD5(),
+                    file(workflow.out.bai[0][1]).name,
+                    workflow.out.versions
+                    ).match() 
+                }
+            )
+        }
+    }
+
+    test("sarscov2 - fastq.gz - paired end") {
+
+        when {
+            workflow {
+                """
+                input[0] = Channel.of([[ id:'test', single_end:false ],
+                        [
+                        file(params.modules_testdata_base_path + 'genomics/sarscov2/illumina/fastq/test_1.fastq.gz', checkIfExists: true),
+                        file(params.modules_testdata_base_path + 'genomics/sarscov2/illumina/fastq/test_2.fastq.gz', checkIfExists: true)
+                        ]
+                        ])
+                input[1] = Channel.value([
+                        [id: 'reference'],
+                        file(params.modules_testdata_base_path + 'genomics/sarscov2/genome/genome.fasta', checkIfExists: true)
+                        ])
+                input[2] = BWA_INDEX.out.index
+                input[3] = Channel.value([
+                        [id: 'intervals'],
+                        file(params.modules_testdata_base_path + 'genomics/sarscov2/genome/picard/baits.interval_list', checkIfExists: true)
+                        ])
+                input[4] = file(params.modules_testdata_base_path + 'genomics/sarscov2/illumina/vcf/test.vcf.gz', checkIfExists: true)
+                """
+            }
+        }
+
+        then {
+            assertAll(
+                { assert workflow.success},
+                { assert snapshot(
+                    bam(workflow.out.bam[0][1]).getReadsMD5(),
+                    file(workflow.out.bai[0][1]).name,
+                    workflow.out.versions
+                    ).match() 
+                }
+            )
+        }
+    }
+}

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

@@ -0,0 +1,137 @@
+{
+    "fastq_align_parabricks_single_end": {
+        "content": [
+            {
+                "0": [
+
+                ],
+                "1": [
+
+                ],
+                "2": [
+
+                ],
+                "bai": [
+
+                ],
+                "bam": [
+
+                ],
+                "versions": [
+
+                ]
+            }
+        ],
+        "meta": {
+            "nf-test": "0.9.1",
+            "nextflow": "24.04.4"
+        },
+        "timestamp": "2024-10-28T16:07:36.691396"
+    },
+    "sarscov2 - fastq.gz - single end": {
+        "content": [
+            {
+                "0": [
+                    [
+                        {
+                            "id": "test",
+                            "single_end": true
+                        },
+                        "test.bam:md5,a989684840fb07e1cf8c74cc2208b283"
+                    ]
+                ],
+                "1": [
+                    [
+                        {
+                            "id": "test",
+                            "single_end": true
+                        },
+                        "test.bam.bai:md5,b42d497c4b8bfda390bc49777fafee75"
+                    ]
+                ],
+                "2": [
+                    "versions.yml:md5,4d671c4d60b6a0279cfca507525daa77"
+                ],
+                "bai": [
+                    [
+                        {
+                            "id": "test",
+                            "single_end": true
+                        },
+                        "test.bam.bai:md5,b42d497c4b8bfda390bc49777fafee75"
+                    ]
+                ],
+                "bam": [
+                    [
+                        {
+                            "id": "test",
+                            "single_end": true
+                        },
+                        "test.bam:md5,a989684840fb07e1cf8c74cc2208b283"
+                    ]
+                ],
+                "versions": [
+                    "versions.yml:md5,4d671c4d60b6a0279cfca507525daa77"
+                ]
+            }
+        ],
+        "meta": {
+            "nf-test": "0.9.2",
+            "nextflow": "24.10.0"
+        },
+        "timestamp": "2024-11-18T11:42:27.869160615"
+    },
+    "sarscov2 - fastq.gz - paired end": {
+        "content": [
+            {
+                "0": [
+                    [
+                        {
+                            "id": "test",
+                            "single_end": false
+                        },
+                        "test.bam:md5,729ae53d6dcb627b478d5e3aa454dd90"
+                    ]
+                ],
+                "1": [
+                    [
+                        {
+                            "id": "test",
+                            "single_end": false
+                        },
+                        "test.bam.bai:md5,ad5084ca0975b685e0f36322ca2fa137"
+                    ]
+                ],
+                "2": [
+                    "versions.yml:md5,4d671c4d60b6a0279cfca507525daa77"
+                ],
+                "bai": [
+                    [
+                        {
+                            "id": "test",
+                            "single_end": false
+                        },
+                        "test.bam.bai:md5,ad5084ca0975b685e0f36322ca2fa137"
+                    ]
+                ],
+                "bam": [
+                    [
+                        {
+                            "id": "test",
+                            "single_end": false
+                        },
+                        "test.bam:md5,729ae53d6dcb627b478d5e3aa454dd90"
+                    ]
+                ],
+                "versions": [
+                    "versions.yml:md5,4d671c4d60b6a0279cfca507525daa77"
+                ]
+            }
+        ],
+        "meta": {
+            "nf-test": "0.9.2",
+            "nextflow": "24.10.0"
+        },
+        "timestamp": "2024-11-18T11:43:11.246974608"
+    }
+}