Merge pull request #1541 from matthdsm/subwf/bam_qc

New subworkflow: bam_qc_picard
2024-12-22 11:08:17 +00:00 · 2022-04-29 14:31:38 +02:00 · 2022-04-29 14:31:38 +02:00 · 0ec3dea37f
commit 0ec3dea37f
parent c55032d10e 7e391c3a3b
5 changed files with 166 additions and 0 deletions
--- a/subworkflows/nf-core/bam_qc_picard/main.nf
+++ b/subworkflows/nf-core/bam_qc_picard/main.nf
@ -0,0 +1,41 @@
+//
+// Run QC steps on BAM/CRAM files using Picard
+//
+
+include { PICARD_COLLECTMULTIPLEMETRICS } from '../../../modules/picard/collectmultiplemetrics/main'
+include { PICARD_COLLECTWGSMETRICS      } from '../../../modules/picard/collectwgsmetrics/main'
+include { PICARD_COLLECTHSMETRICS       } from '../../../modules/picard/collecthsmetrics/main'
+
+workflow BAM_QC_PICARD {
+    take:
+    ch_bam              // channel: [ val(meta), [ bam ]]
+    ch_fasta            // channel: [ fasta ]
+    ch_fasta_fai        // channel: [ fasta_fai ]
+    ch_bait_interval    // channel: [ bait_interval ]
+    ch_target_interval  // channel: [ target_interval ]
+
+    main:
+    ch_versions = Channel.empty()
+    ch_coverage_metrics = Channel.empty()
+
+    PICARD_COLLECTMULTIPLEMETRICS( ch_bam, ch_fasta )
+    ch_versions = ch_versions.mix(PICARD_COLLECTMULTIPLEMETRICS.out.versions.first())
+
+    if (ch_bait_interval || ch_target_interval) {
+        if (!ch_bait_interval) log.error("Bait interval channel is empty")
+        if (!ch_target_interval) log.error("Target interval channel is empty")
+        PICARD_COLLECTHSMETRICS( ch_bam, ch_fasta, ch_fasta_fai, ch_bait_interval, ch_target_interval )
+        ch_coverage_metrics = ch_coverage_metrics.mix(PICARD_COLLECTHSMETRICS.out.metrics)
+        ch_versions = ch_versions.mix(PICARD_COLLECTHSMETRICS.out.versions.first())
+    } else {
+        PICARD_COLLECTWGSMETRICS( ch_bam, ch_fasta )
+        ch_versions = ch_versions.mix(PICARD_COLLECTWGSMETRICS.out.versions.first())
+        ch_coverage_metrics = ch_coverage_metrics.mix(PICARD_COLLECTWGSMETRICS.out.metrics)
+    }
+
+    emit:
+    coverage_metrics    = ch_coverage_metrics                       // channel: [ val(meta), [ coverage_metrics ] ]
+    multiple_metrics    = PICARD_COLLECTMULTIPLEMETRICS.out.metrics // channel: [ val(meta), [ multiple_metrics ] ]
+
+    versions            = ch_versions                               // channel: [ versions.yml ]
+}
--- a/subworkflows/nf-core/bam_qc_picard/meta.yml
+++ b/subworkflows/nf-core/bam_qc_picard/meta.yml
@ -0,0 +1,60 @@
+name: bam_qc
+description: Produces comprehensive statistics from BAM file
+keywords:
+  - statistics
+  - counts
+  - hs_metrics
+  - wgs_metrics
+  - bam
+  - sam
+  - cram
+modules:
+  - picard/collectmultiplemetrics
+  - picard/collectwgsmetrics
+  - picard/collecthsmetrics
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - bam:
+      type: file
+      description: BAM/CRAM/SAM file
+      pattern: "*.{bam,cram,sam}"
+  - fasta:
+      type: optional file
+      description: Reference fasta file
+      pattern: "*.{fasta,fa}"
+  - fasta_fai:
+      type: optional file
+      description: Reference fasta file index
+      pattern: "*.{fasta,fa}.fai"
+  - bait_intervals:
+      type: optional file
+      description: An interval list file that contains the locations of the baits used.
+      pattern: "baits.interval_list"
+  - target_intervals:
+      type: optional file
+      description: An interval list file that contains the locations of the targets.
+      pattern: "targets.interval_list"
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - coverage_metrics:
+      type: file
+      description: Alignment metrics files generated by picard CollectHsMetrics or CollectWgsMetrics
+      pattern: "*_metrics.txt"
+  - multiple_metrics:
+      type: file
+      description: Alignment metrics files generated by picard CollectMultipleMetrics
+      pattern: "*_{metrics}"
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+authors:
+  - "@matthdsm"
--- a/tests/subworkflows/nf-core/bam_qc_picard/main.nf
+++ b/tests/subworkflows/nf-core/bam_qc_picard/main.nf
@ -0,0 +1,27 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { BAM_QC_PICARD } from '../../../../subworkflows/nf-core/bam_qc_picard/main' addParams([:])
+
+workflow test_bam_qc_picard_wgs {
+    input = [ [ id:'test', single_end:false ], // meta map
+                file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true)
+            ]
+    fasta   = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+    fasta_fai   = file(params.test_data['sarscov2']['genome']['genome_fasta_fai'], checkIfExists: true)
+
+    BAM_QC_PICARD ( input, fasta, fasta_fai, [], [] )
+}
+
+workflow test_bam_qc_picard_targetted {
+    input = [ [ id:'test', single_end:false ], // meta map
+                file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true)
+            ]
+    fasta       = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+    fasta_fai   = file(params.test_data['sarscov2']['genome']['genome_fasta_fai'], checkIfExists: true)
+    bait        = file(params.test_data['sarscov2']['genome']['baits_interval_list'], checkIfExists: true)
+    target      = file(params.test_data['sarscov2']['genome']['targets_interval_list'], checkIfExists: true)
+
+    BAM_QC_PICARD ( input, fasta, fasta_fai, bait, target )
+}
--- a/tests/subworkflows/nf-core/bam_qc_picard/nextflow.config
+++ b/tests/subworkflows/nf-core/bam_qc_picard/nextflow.config
@ -0,0 +1,5 @@
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+
+}
--- a/tests/subworkflows/nf-core/bam_qc_picard/test.yml
+++ b/tests/subworkflows/nf-core/bam_qc_picard/test.yml
@ -0,0 +1,33 @@
+- name: bam qc picard wgs
+  command: nextflow run ./tests/subworkflows/nf-core/bam_qc_picard -entry test_bam_qc_picard_wgs -c tests/config/nextflow.config
+  tags:
+    - subworkflows
+    # - subworkflows/bam_qc_picard
+    # Modules
+    # - picard
+    # - picard/collectmultiplemetrics
+    # - picard/collectwgsmetrics
+  files:
+    - path: ./output/picard/test.CollectMultipleMetrics.alignment_summary_metrics
+    - path: ./output/picard/test.CollectMultipleMetrics.insert_size_metrics
+    - path: ./output/picard/test.CollectMultipleMetrics.base_distribution_by_cycle_metrics
+    - path: ./output/picard/test.CollectMultipleMetrics.quality_by_cycle_metrics
+    - path: ./output/picard/test.CollectMultipleMetrics.quality_distribution_metrics
+    - path: ./output/picard/test.CollectWgsMetrics.coverage_metrics
+
+- name: bam qc picard targetted
+  command: nextflow run ./tests/subworkflows/nf-core/bam_qc_picard -entry test_bam_qc_picard_targetted -c tests/config/nextflow.config
+  tags:
+    - subworkflows
+    # - subworkflows/bam_qc_picard
+    # Modules
+    # - picard
+    # - picard/collectmultiplemetrics
+    # - picard/collecthsmetrics
+  files:
+    - path: ./output/picard/test.CollectMultipleMetrics.alignment_summary_metrics
+    - path: ./output/picard/test.CollectMultipleMetrics.insert_size_metrics
+    - path: ./output/picard/test.CollectMultipleMetrics.base_distribution_by_cycle_metrics
+    - path: ./output/picard/test.CollectMultipleMetrics.quality_by_cycle_metrics
+    - path: ./output/picard/test.CollectMultipleMetrics.quality_distribution_metrics
+    - path: ./output/picard/test.CollectHsMetrics.coverage_metrics