Merge pull request #1701 from matthdsm/update/mosdepth

Mosdepth: add all remaining outputs
2024-12-22 11:08:17 +00:00 · 2022-05-25 19:46:37 +01:00 · 2022-05-25 19:46:37 +01:00 · f89b76afa4
commit f89b76afa4
parent 038fced07e 72a31b76eb
5 changed files with 146 additions and 48 deletions
--- a/modules/mosdepth/main.nf
+++ b/modules/mosdepth/main.nf
@ -14,13 +14,17 @@ process MOSDEPTH {

    output:
    tuple val(meta), path('*.global.dist.txt')      , emit: global_txt
-    tuple val(meta), path('*.region.dist.txt')    , emit: regions_txt , optional:true
    tuple val(meta), path('*.summary.txt')          , emit: summary_txt
-    tuple val(meta), path('*.per-base.d4')        , emit: per_base_d4 , optional:true
-    tuple val(meta), path('*.per-base.bed.gz')    , emit: per_base_bed, optional:true
-    tuple val(meta), path('*.per-base.bed.gz.csi'), emit: per_base_csi, optional:true
-    tuple val(meta), path('*.regions.bed.gz')     , emit: regions_bed , optional:true
-    tuple val(meta), path('*.regions.bed.gz.csi') , emit: regions_csi , optional:true
+    tuple val(meta), path('*.region.dist.txt')      , optional:true, emit: regions_txt
+    tuple val(meta), path('*.per-base.d4')          , optional:true, emit: per_base_d4
+    tuple val(meta), path('*.per-base.bed.gz')      , optional:true, emit: per_base_bed
+    tuple val(meta), path('*.per-base.bed.gz.csi')  , optional:true, emit: per_base_csi
+    tuple val(meta), path('*.regions.bed.gz')       , optional:true, emit: regions_bed
+    tuple val(meta), path('*.regions.bed.gz.csi')   , optional:true, emit: regions_csi
+    tuple val(meta), path('*.quantized.bed.gz')     , optional:true, emit: quantized_bed
+    tuple val(meta), path('*.quantized.bed.gz.csi') , optional:true, emit: quantized_csi
+    tuple val(meta), path('*.thresholds.bed.gz')    , optional:true, emit: thresholds_bed
+    tuple val(meta), path('*.thresholds.bed.gz.csi'), optional:true, emit: thresholds_csi
    path  "versions.yml"                            , emit: versions

    when:
@ -34,10 +38,13 @@ process MOSDEPTH {
    if (bed && args.contains("--by")) {
        exit 1, "'--by' can only be specified once when running mosdepth! Either remove input BED file definition or remove '--by' from 'ext.args' definition"
    }
+    if (!bed && args.contains("--thresholds")) {
+        exit 1, "'--thresholds' can only be specified in conjunction with '--by'"
+    }

    """
    mosdepth \\
-        --threads ${task.cpus} \\
+        --threads $task.cpus \\
        $interval \\
        $reference \\
        $args \\
@ -61,6 +68,10 @@ process MOSDEPTH {
    touch ${prefix}.per-base.bed.gz.csi
    touch ${prefix}.regions.bed.gz
    touch ${prefix}.regions.bed.gz.csi
+    touch ${prefix}.quantized.bed.gz
+    touch ${prefix}.quantized.bed.gz.csi
+    touch ${prefix}.thresholds.bed.gz
+    touch ${prefix}.thresholds.bed.gz.csi

    cat <<-END_VERSIONS > versions.yml
    "${task.process}":
--- a/modules/mosdepth/meta.yml
+++ b/modules/mosdepth/meta.yml
@ -72,6 +72,22 @@ output:
      type: file
      description: Index file for BED file with per-region coverage
      pattern: "*.{regions.bed.gz.csi}"
+  - quantized_bed:
+      type: file
+      description: BED file with binned coverage
+      pattern: "*.{quantized.bed.gz}"
+  - quantized_csi:
+      type: file
+      description: Index file for BED file with binned coverage
+      pattern: "*.{quantized.bed.gz.csi}"
+  - thresholds_bed:
+      type: file
+      description: BED file with the number of bases in each region that are covered at or above each threshold
+      pattern: "*.{thresholds.bed.gz}"
+  - thresholds_csi:
+      type: file
+      description: Index file for BED file with threshold coverage
+      pattern: "*.{thresholds.bed.gz.csi}"
  - versions:
      type: file
      description: File containing software versions
--- a/tests/modules/mosdepth/main.nf
+++ b/tests/modules/mosdepth/main.nf
@ -5,12 +5,14 @@ nextflow.enable.dsl = 2
 include { MOSDEPTH                       } from '../../../modules/mosdepth/main.nf'
 include { MOSDEPTH as MOSDEPTH_FAIL      } from '../../../modules/mosdepth/main.nf'
 include { MOSDEPTH as MOSDEPTH_WINDOW    } from '../../../modules/mosdepth/main.nf'
+include { MOSDEPTH as MOSDEPTH_THRESHOLD } from '../../../modules/mosdepth/main.nf'
+include { MOSDEPTH as MOSDEPTH_QUANTIZED } from '../../../modules/mosdepth/main.nf'

 workflow test_mosdepth {
    input = [
        [ id:'test', single_end:true ],
-                [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true) ],
-                [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true) ]
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true)
    ]

    MOSDEPTH ( input, [], [] )
@ -19,10 +21,10 @@ workflow test_mosdepth {
 workflow test_mosdepth_bed {
    input = [
        [ id:'test', single_end:true ],
-                [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true) ],
-                [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true) ]
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true)
    ]
-    bed  = [ file(params.test_data['homo_sapiens']['genome']['genome_bed'], checkIfExists: true) ]
+    bed = file(params.test_data['homo_sapiens']['genome']['genome_bed'], checkIfExists: true)

    MOSDEPTH ( input, bed, [] )
 }
@ -30,10 +32,10 @@ workflow test_mosdepth_bed {
 workflow test_mosdepth_cram {
    input = [
        [ id:'test', single_end:true ],
-                [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_cram'], checkIfExists: true) ],
-                [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_cram_crai'], checkIfExists: true) ]
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_cram'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_cram_crai'], checkIfExists: true)
    ]
-    fasta = [ file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true) ]
+    fasta = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)

    MOSDEPTH ( input, [], fasta )
 }
@ -41,11 +43,11 @@ workflow test_mosdepth_cram {
 workflow test_mosdepth_cram_bed {
    input  = [
        [ id:'test', single_end:true ],
-                [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_cram'], checkIfExists: true) ],
-                [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_cram_crai'], checkIfExists: true) ]
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_cram'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_cram_crai'], checkIfExists: true)
    ]
-    bed  = [ file(params.test_data['homo_sapiens']['genome']['genome_bed'], checkIfExists: true) ]
-    fasta = [ file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true) ]
+    bed = file(params.test_data['homo_sapiens']['genome']['genome_bed'], checkIfExists: true)
+    fasta = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)

    MOSDEPTH ( input, bed, fasta )
 }
@ -53,21 +55,42 @@ workflow test_mosdepth_cram_bed {
 workflow test_mosdepth_window {
    input = [
        [ id:'test', single_end:true ],
-                [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true) ],
-                [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true) ]
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true)
    ]
-    bed  = [ file(params.test_data['homo_sapiens']['genome']['genome_bed'], checkIfExists: true) ]
+    bed = file(params.test_data['homo_sapiens']['genome']['genome_bed'], checkIfExists: true)

    MOSDEPTH_WINDOW ( input, [], [] )
 }

+workflow test_mosdepth_quantized {
+    input = [
+        [ id:'test', single_end:true ],
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true)
+    ]
+
+    MOSDEPTH_QUANTIZED ( input, [], [] )
+}
+
+workflow test_mosdepth_thresholds {
+    input = [
+        [ id:'test', single_end:true ],
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true)
+    ]
+    bed = file(params.test_data['homo_sapiens']['genome']['genome_bed'], checkIfExists: true)
+
+    MOSDEPTH_THRESHOLD ( input, bed, [] )
+}
+
 workflow test_mosdepth_fail {
    input = [
        [ id:'test', single_end:true ],
-                [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true) ],
-                [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true) ]
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true)
    ]
-    bed  = [ file(params.test_data['homo_sapiens']['genome']['genome_bed'], checkIfExists: true) ]
+    bed = file(params.test_data['homo_sapiens']['genome']['genome_bed'], checkIfExists: true)

    MOSDEPTH_FAIL ( input, bed, [] )
 }
--- a/tests/modules/mosdepth/nextflow.config
+++ b/tests/modules/mosdepth/nextflow.config
@ -7,4 +7,10 @@ process {
    withName: MOSDEPTH_WINDOW {
        ext.args = "--by 100"
    }
+    withName: MOSDEPTH_QUANTIZED {
+        ext.args = "--quantize 0:1:4:100:200"
+    }
+    withName: MOSDEPTH_THRESHOLD {
+        ext.args = "--thresholds 1,10,20,30"
+    }
 }
--- a/tests/modules/mosdepth/test.yml
+++ b/tests/modules/mosdepth/test.yml
@ -86,6 +86,48 @@
    - path: output/mosdepth/test.regions.bed.gz.csi
      md5sum: 257d67678136963d9dd904330079609d

+- name: mosdepth test_mosdepth_quantized
+  command: nextflow run ./tests/modules/mosdepth -entry test_mosdepth_quantized -c ./tests/config/nextflow.config  -c ./tests/modules/mosdepth/nextflow.config
+  tags:
+    - mosdepth
+  files:
+    - path: output/mosdepth/test.mosdepth.global.dist.txt
+      md5sum: e82e90c7d508a135b5a8a7cd6933452e
+    - path: output/mosdepth/test.mosdepth.summary.txt
+      md5sum: 4f0d231060cbde4efdd673863bd2fb59
+    - path: output/mosdepth/test.per-base.bed.gz
+      md5sum: bc1df47d46f818fee5275975925d769a
+    - path: output/mosdepth/test.per-base.bed.gz.csi
+      md5sum: 9e649ac749ff6c6073bef5ab63e8aaa4
+    - path: output/mosdepth/test.quantized.bed.gz
+      md5sum: 3e434a8bafcf59a67841ae3d4d752838
+    - path: output/mosdepth/test.quantized.bed.gz.csi
+      md5sum: be9617f551f19a33923f1e886eaefb93
+
+- name: mosdepth test_mosdepth_thresholds
+  command: nextflow run ./tests/modules/mosdepth -entry test_mosdepth_thresholds -c ./tests/config/nextflow.config  -c ./tests/modules/mosdepth/nextflow.config
+  tags:
+    - mosdepth
+  files:
+    - path: output/mosdepth/test.mosdepth.global.dist.txt
+      md5sum: e82e90c7d508a135b5a8a7cd6933452e
+    - path: output/mosdepth/test.mosdepth.region.dist.txt
+      md5sum: e82e90c7d508a135b5a8a7cd6933452e
+    - path: output/mosdepth/test.mosdepth.summary.txt
+      md5sum: 96c037f769974b904beb53edc4f56d82
+    - path: output/mosdepth/test.per-base.bed.gz
+      md5sum: bc1df47d46f818fee5275975925d769a
+    - path: output/mosdepth/test.per-base.bed.gz.csi
+      md5sum: 9e649ac749ff6c6073bef5ab63e8aaa4
+    - path: output/mosdepth/test.regions.bed.gz
+      md5sum: 5d398caf7171ec4406278e2add3009ae
+    - path: output/mosdepth/test.regions.bed.gz.csi
+      md5sum: 47669cfe41f3e222e74d81e1b1be191f
+    - path: output/mosdepth/test.thresholds.bed.gz
+      md5sum: 13101e326eea3cbfa1d569b69f494f4c
+    - path: output/mosdepth/test.thresholds.bed.gz.csi
+      md5sum: 912055ee9452229439df6fae95644196
+
 - name: mosdepth test_mosdepth_fail
  command: nextflow run ./tests/modules/mosdepth -entry test_mosdepth_fail -c ./tests/config/nextflow.config  -c ./tests/modules/mosdepth/nextflow.config
  tags: