Merge branch 'master' into cnvkit/reference

2024-12-22 11:08:17 +00:00 · 2022-05-24 15:00:29 +02:00 · 2022-05-24 15:00:29 +02:00 · 3077e4d34c
commit 3077e4d34c
parent 8208140d21 2c0b0df088
5 changed files with 166 additions and 77 deletions
--- a/modules/mosdepth/main.nf
+++ b/modules/mosdepth/main.nf
@ -10,13 +10,13 @@ process MOSDEPTH {
    input:
    tuple val(meta), path(bam), path(bai)
    path  bed
-    val   window_size
+    path  fasta

    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: d4          , optional:true
+    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
@ -29,19 +29,21 @@ process MOSDEPTH {
    script:
    def args = task.ext.args ?: ''
    def prefix = task.ext.prefix ?: "${meta.id}"
-    if (window_size) {
-        interval = "--by ${window_size}"
-    } else if ( bed ) {
-        interval = "--by ${bed}"
-    } else {
-        interval = ""
+    def reference = fasta ? "--fasta ${fasta}" : ""
+    def interval = bed ? "--by ${bed}" : ""
+    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"
    }
+
    """
    mosdepth \\
+        --threads ${task.cpus} \\
        $interval \\
+        $reference \\
        $args \\
        $prefix \\
        $bam
+
    cat <<-END_VERSIONS > versions.yml
    "${task.process}":
        mosdepth: \$(mosdepth --version 2>&1 | sed 's/^.*mosdepth //; s/ .*\$//')
--- a/modules/mosdepth/meta.yml
+++ b/modules/mosdepth/meta.yml
@ -30,10 +30,10 @@ input:
      type: file
      description: BED file with intersected intervals
      pattern: "*.{bed}"
-  - window_size:
-      type: integer
-      description: Window size
-      pattern: "[0-9]+"
+  - fasta:
+      type: file
+      description: Reference genome FASTA file
+      pattern: "*.{fa,fasta}"
 output:
  - meta:
      type: map
@ -60,6 +60,10 @@ output:
      type: file
      description: Index file for BED file with per-base coverage
      pattern: "*.{per-base.bed.gz.csi}"
+  - per_base_d4:
+      type: file
+      description: D4 file with per-base coverage
+      pattern: "*.{per-base.d4}"
  - regions_bed:
      type: file
      description: BED file with per-region coverage
@ -76,3 +80,4 @@ authors:
  - "@joseespinosa"
  - "@drpatelh"
  - "@ramprasadn"
+  - "@matthdsm"
--- a/tests/modules/mosdepth/main.nf
+++ b/tests/modules/mosdepth/main.nf
@ -3,34 +3,71 @@
 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'

 workflow test_mosdepth {
-    input  = [ [ id:'test', single_end:true ],
-               [ file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true) ],
-               [ file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true) ]
+    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 ( input, [], [] )
 }

-
-workflow test_mosdepth_window {
-    input  = [ [ id:'test', single_end:true ],
-               [ file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true) ],
-               [ file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true) ]
-             ]
-    window = 100
-
-    MOSDEPTH ( input, [], window )
-}
-
-
 workflow test_mosdepth_bed {
-    input  = [ [ id:'test', single_end:true ],
-               [ file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true) ],
-               [ file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true) ]
+    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['sarscov2']['genome']['test_bed'], checkIfExists: true) ]
+    bed  = [ file(params.test_data['homo_sapiens']['genome']['genome_bed'], checkIfExists: true) ]

    MOSDEPTH ( input, 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) ]
+            ]
+    fasta = [ file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true) ]
+
+    MOSDEPTH ( input, [], fasta )
+}
+
+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) ]
+            ]
+    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 )
+}
+
+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) ]
+            ]
+    bed  = [ file(params.test_data['homo_sapiens']['genome']['genome_bed'], checkIfExists: true) ]
+
+    MOSDEPTH_WINDOW ( input, [], [] )
+}
+
+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) ]
+            ]
+    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
@ -1,5 +1,10 @@
 process {

    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
-
+    withName: MOSDEPTH_FAIL {
+        ext.args = "--by 100"
+    }
+    withName: MOSDEPTH_WINDOW {
+        ext.args = "--by 100"
+    }
 }
--- a/tests/modules/mosdepth/test.yml
+++ b/tests/modules/mosdepth/test.yml
@ -1,53 +1,93 @@
- name: mosdepth
+- name: mosdepth test_mosdepth
  command: nextflow run ./tests/modules/mosdepth -entry test_mosdepth -c ./tests/config/nextflow.config  -c ./tests/modules/mosdepth/nextflow.config
  tags:
    - mosdepth
  files:
-    - path: ./output/mosdepth/test.per-base.bed.gz.csi
-      md5sum: b2aad62c41a7146680d31df505fcc8c5
-    - path: ./output/mosdepth/test.per-base.bed.gz
-      md5sum: 11b3f649072c2c7453febb085b1a9c33
-    - path: ./output/mosdepth/test.mosdepth.global.dist.txt
-      md5sum: 2a1de1b0ecc361a21cd296ec4e1efd6a
-    - path: ./output/mosdepth/test.mosdepth.summary.txt
-      md5sum: 7b249dd3b3e58cc122fbd25ea84aa25d
+    - 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

- name: mosdepth window
-  command: nextflow run ./tests/modules/mosdepth -entry test_mosdepth_window -c ./tests/config/nextflow.config -c ./tests/modules/mosdepth/nextflow.config
-  tags:
-    - mosdepth
-  files:
-    - path: ./output/mosdepth/test.per-base.bed.gz.csi
-      md5sum: b2aad62c41a7146680d31df505fcc8c5
-    - path: ./output/mosdepth/test.per-base.bed.gz
-      md5sum: 11b3f649072c2c7453febb085b1a9c33
-    - path: ./output/mosdepth/test.mosdepth.global.dist.txt
-      md5sum: 2a1de1b0ecc361a21cd296ec4e1efd6a
-    - path: ./output/mosdepth/test.regions.bed.gz
-      md5sum: 64e1ced01c4443d7c1796ef553992f0c
-    - path: ./output/mosdepth/test.regions.bed.gz.csi
-      md5sum: 9e312b4b0784bd46dfbd23b3a8afed6a
-    - path: ./output/mosdepth/test.mosdepth.region.dist.txt
-      md5sum: 65fbc824c4212c6884354d8ac72ad37e
-    - path: ./output/mosdepth/test.mosdepth.summary.txt
-      md5sum: 11804907dab069ddb99ca97bf2698572
-
- name: mosdepth bed
+- name: mosdepth test_mosdepth_bed
  command: nextflow run ./tests/modules/mosdepth -entry test_mosdepth_bed -c ./tests/config/nextflow.config  -c ./tests/modules/mosdepth/nextflow.config
  tags:
    - mosdepth
  files:
-    - path: ./output/mosdepth/test.per-base.bed.gz.csi
-      md5sum: b2aad62c41a7146680d31df505fcc8c5
-    - path: ./output/mosdepth/test.per-base.bed.gz
-      md5sum: 11b3f649072c2c7453febb085b1a9c33
-    - path: ./output/mosdepth/test.mosdepth.global.dist.txt
-      md5sum: 2a1de1b0ecc361a21cd296ec4e1efd6a
-    - path: ./output/mosdepth/test.regions.bed.gz
-      md5sum: 347f877700d1dc42c95157199eff25d5
-    - path: ./output/mosdepth/test.regions.bed.gz.csi
-      md5sum: ed5fbf46e3bdcbf60094df295bc40356
-    - path: ./output/mosdepth/test.mosdepth.region.dist.txt
-      md5sum: 295564628113d2ec0ca34d7f661cfea8
-    - path: ./output/mosdepth/test.mosdepth.summary.txt
-      md5sum: b07817412fd17819c14541e63bc4926c
+    - 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
+
+- name: mosdepth test_mosdepth_cram
+  command: nextflow run ./tests/modules/mosdepth -entry test_mosdepth_cram -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
+
+- name: mosdepth test_mosdepth_cram_bed
+  command: nextflow run ./tests/modules/mosdepth -entry test_mosdepth_cram_bed -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
+
+- name: mosdepth test_mosdepth_window
+  command: nextflow run ./tests/modules/mosdepth -entry test_mosdepth_window -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: 39e0e707ec32feb5176fd20a95f1f468
+    - 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: f02e2cb49cc050e13d76942d6960827a
+    - path: output/mosdepth/test.regions.bed.gz.csi
+      md5sum: 257d67678136963d9dd904330079609d
+
+- 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:
+    - mosdepth
+  exit_code: 1