New sbwf bam_subsampledepth_samtools

subworkflows/nf-core/bam_subsampledepth_samtools/main.nf

@@ -0,0 +1,51 @@
+include { SAMTOOLS_DEPTH } from '../../../modules/nf-core/samtools/depth'
+include { GAWK           } from '../../../modules/nf-core/gawk'
+include { SAMTOOLS_VIEW  } from '../../../modules/nf-core/samtools/view'
+
+workflow BAM_SUBSAMPLEDEPTH_SAMTOOLS {
+
+    take:
+    ch_bam_bai_depth    // channel: [ val(meta), path(bam), path(bai), val(depth) ]
+    ch_fasta            // channel: [ val(meta), path(fasta) ]
+
+    main:
+    ch_versions      = Channel.empty()
+
+    // Compute mean depth
+    SAMTOOLS_DEPTH(ch_bam_bai_depth.map{ it[0..2] }, [[], []])
+    ch_versions = ch_versions.mix(SAMTOOLS_DEPTH.out.versions.first())
+
+    // Use GAWK to get mean depth
+    GAWK(SAMTOOLS_DEPTH.out.tsv, [])
+    ch_versions = ch_versions.mix(GAWK.out.versions.first())
+
+    // Compute downsampling factor
+    ch_mean_depth = GAWK.out.output
+        .splitCsv(header: false, sep:'\t')
+        .map{ meta, row ->
+            [ meta, row[0] as Float ]
+        }
+
+    // Add all necessary channel for downsampling
+    ch_input_subsample = ch_bam_bai_depth
+        .join(ch_mean_depth)
+        .map{ meta, bam, index, depth, mean ->
+            [ meta + ['subsample_fraction': depth as Float / mean, 'depth': depth ], bam, index ]
+        }
+
+    // Downsample
+    SAMTOOLS_VIEW(
+        ch_input_subsample,
+        ch_fasta,
+        []
+    )
+    ch_versions = ch_versions.mix(SAMTOOLS_VIEW.out.versions.first())
+
+    // Aggregate bam and index
+    ch_bam_subsampled = SAMTOOLS_VIEW.out.bam.mix(SAMTOOLS_VIEW.out.cram, SAMTOOLS_VIEW.out.sam)
+        .join(SAMTOOLS_VIEW.out.bai.mix(SAMTOOLS_VIEW.out.crai, SAMTOOLS_VIEW.out.csi))
+
+    emit:
+    bam_subsampled    = ch_bam_subsampled             // channel: [ val(meta), path(bam), path(csi) ]
+    versions          = ch_versions                   // channel: [ path(versions.yml) ]
+}

subworkflows/nf-core/bam_subsampledepth_samtools/meta.yml

@@ -0,0 +1,42 @@
+# yaml-language-server: $schema=https://raw.githubusercontent.com/nf-core/modules/master/subworkflows/yaml-schema.json
+name: "bam_subsampledepth_samtools"
+description: Subsample a BAM/CRAM/SAM file using samtools to a given mean depth
+keywords:
+  - subsample
+  - bam
+  - sam
+  - cram
+components:
+  - samtools/depth
+  - samtools/view
+  - gawk
+input:
+  - ch_bam:
+      type: file
+      description: |
+        The input channel containing the BAM/CRAM/SAM files and their indexes and the depth at which to subsample them.
+        Structure: [ val(meta), path(bam), path(bai), val(depth) ]
+      pattern: "*.{bam,cram,sam}"
+  - ch_fasta:
+      type: file
+      description: |
+        The reference genome channel containing the fasta files and its index
+        Structure: [ val(meta), path(fasta), path(fai) ]
+      pattern: "*.{fa(sta)?}"
+output:
+  - bam_subsampled:
+      type: file
+      description: |
+        Channel containing subsampled BAM/CRAM/SAM files and their indexes
+        Structure: [ val(meta), path(bam), path(csi) ]
+      pattern: "*.{bam,cram,sam}"
+  - versions:
+      type: file
+      description: |
+        File containing software versions
+        Structure: [ path(versions.yml) ]
+      pattern: "versions.yml"
+authors:
+  - "@louislenezet"
+maintainers:
+  - "@louislenezet"

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

@@ -0,0 +1,52 @@
+nextflow_workflow {
+
+    name "Test Subworkflow BAM_SUBSAMPLEDEPTH_SAMTOOLS"
+    script "../main.nf"
+    config "./nextflow.config"
+
+    workflow "BAM_SUBSAMPLEDEPTH_SAMTOOLS"
+
+    tag "subworkflows"
+    tag "subworkflows_nfcore"
+    tag "subworkflows/bam_subsampledepth_samtools"
+
+    tag "samtools"
+    tag "samtools/depth"
+    tag "samtools/view"
+    tag "gawk"
+
+    test("Downsample to 4X and 2X") {
+        when {
+            workflow {
+                """
+                input[0] = Channel.fromList([
+                    [
+                        [id: "NA12878"],
+                        file("https://raw.githubusercontent.com/nf-core/test-datasets/phaseimpute/hum_data/individuals/NA12878/NA12878.s.bam", checkIfExist:true),
+                        file("https://raw.githubusercontent.com/nf-core/test-datasets/phaseimpute/hum_data/individuals/NA12878/NA12878.s.bam.bai", checkIfExist:true),
+                    ],
+                    [
+                        [id: "NA19401"],
+                        file("https://raw.githubusercontent.com/nf-core/test-datasets/phaseimpute/hum_data/individuals/NA19401/NA19401.s.bam", checkIfExist:true),
+                        file("https://raw.githubusercontent.com/nf-core/test-datasets/phaseimpute/hum_data/individuals/NA19401/NA19401.s.bam.bai", checkIfExist:true),
+                    ],
+                ]). combine( Channel.of(2, 4))
+                input[1] = Channel.of([
+                    [id: "GRCh38"],
+                    file("https://raw.githubusercontent.com/nf-core/test-datasets/phaseimpute/hum_data/reference_genome/GRCh38.s.fa.gz", checkIfExist:true),
+                ]).collect()
+                """
+            }
+        }
+
+        then {
+            assertAll(
+                { assert workflow.success },
+                { assert snapshot(
+                    workflow.out,
+                    workflow.out.bam_subsampled.collect{ [it[0], bam(it[1]).getReads().size()] }
+                ).match() }
+            )
+        }
+    }
+}

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

@@ -0,0 +1,133 @@
+{
+    "Downsample to 4X and 2X": {
+        "content": [
+            {
+                "0": [
+                    [
+                        {
+                            "id": "NA12878",
+                            "subsample_fraction": 0.06201800538763442,
+                            "depth": 2
+                        },
+                        "NA12878.bam:md5,c998482010b83365a4889c3fa75ac578",
+                        "NA12878.bam.csi:md5,0b4abd161cdcc2b51571c9cc651df354"
+                    ],
+                    [
+                        {
+                            "id": "NA12878",
+                            "subsample_fraction": 0.12403601077526884,
+                            "depth": 4
+                        },
+                        "NA12878.bam:md5,d6045df32f7c77d5c863b2068739faa9",
+                        "NA12878.bam.csi:md5,3f8f9a17463b6c2391c632681d3f253a"
+                    ],
+                    [
+                        {
+                            "id": "NA19401",
+                            "subsample_fraction": 0.062137851766009305,
+                            "depth": 2
+                        },
+                        "NA19401.bam:md5,6b144e7adc1f9e6711aea0e4772c5937",
+                        "NA19401.bam.csi:md5,36c29896003f2de5306ff427a969058c"
+                    ],
+                    [
+                        {
+                            "id": "NA19401",
+                            "subsample_fraction": 0.12427570353201861,
+                            "depth": 4
+                        },
+                        "NA19401.bam:md5,3b0d20e5f44952135a547c6230117460",
+                        "NA19401.bam.csi:md5,b44701d7d4de827cc1b83aaeb397deaf"
+                    ]
+                ],
+                "1": [
+                    "versions.yml:md5,8019e4c6fa3c1ddea25d64d6bfe5651f",
+                    "versions.yml:md5,b31618773ed8a31f8635bca3da001eeb",
+                    "versions.yml:md5,f134df55b5047c5a8222ca04cf5ec501"
+                ],
+                "bam_subsampled": [
+                    [
+                        {
+                            "id": "NA12878",
+                            "subsample_fraction": 0.06201800538763442,
+                            "depth": 2
+                        },
+                        "NA12878.bam:md5,c998482010b83365a4889c3fa75ac578",
+                        "NA12878.bam.csi:md5,0b4abd161cdcc2b51571c9cc651df354"
+                    ],
+                    [
+                        {
+                            "id": "NA12878",
+                            "subsample_fraction": 0.12403601077526884,
+                            "depth": 4
+                        },
+                        "NA12878.bam:md5,d6045df32f7c77d5c863b2068739faa9",
+                        "NA12878.bam.csi:md5,3f8f9a17463b6c2391c632681d3f253a"
+                    ],
+                    [
+                        {
+                            "id": "NA19401",
+                            "subsample_fraction": 0.062137851766009305,
+                            "depth": 2
+                        },
+                        "NA19401.bam:md5,6b144e7adc1f9e6711aea0e4772c5937",
+                        "NA19401.bam.csi:md5,36c29896003f2de5306ff427a969058c"
+                    ],
+                    [
+                        {
+                            "id": "NA19401",
+                            "subsample_fraction": 0.12427570353201861,
+                            "depth": 4
+                        },
+                        "NA19401.bam:md5,3b0d20e5f44952135a547c6230117460",
+                        "NA19401.bam.csi:md5,b44701d7d4de827cc1b83aaeb397deaf"
+                    ]
+                ],
+                "versions": [
+                    "versions.yml:md5,8019e4c6fa3c1ddea25d64d6bfe5651f",
+                    "versions.yml:md5,b31618773ed8a31f8635bca3da001eeb",
+                    "versions.yml:md5,f134df55b5047c5a8222ca04cf5ec501"
+                ]
+            },
+            [
+                [
+                    {
+                        "id": "NA12878",
+                        "subsample_fraction": 0.06201800538763442,
+                        "depth": 2
+                    },
+                    1164
+                ],
+                [
+                    {
+                        "id": "NA12878",
+                        "subsample_fraction": 0.12403601077526884,
+                        "depth": 4
+                    },
+                    2402
+                ],
+                [
+                    {
+                        "id": "NA19401",
+                        "subsample_fraction": 0.062137851766009305,
+                        "depth": 2
+                    },
+                    1196
+                ],
+                [
+                    {
+                        "id": "NA19401",
+                        "subsample_fraction": 0.12427570353201861,
+                        "depth": 4
+                    },
+                    2321
+                ]
+            ]
+        ],
+        "meta": {
+            "nf-test": "0.8.4",
+            "nextflow": "24.04.3"
+        },
+        "timestamp": "2024-07-24T11:40:16.846985786"
+    }
+}

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

@@ -0,0 +1,9 @@
+process {
+    withName: GAWK {
+        ext.args2 = "'{ total += \$3 } END { print total/NR }'"
+        ext.suffix = "txt"
+    }
+    withName: SAMTOOLS_VIEW {
+        ext.args = { "--write-index --subsample ${meta.subsample_fraction}" }
+    }
+}

subworkflows/nf-core/bam_subsampledepth_samtools/tests/tags.yml

@@ -0,0 +1,2 @@
+subworkflows/bam_subsampledepth_samtools:
+  - subworkflows/nf-core/bam_subsampledepth_samtools/**