bedtools/genomecov updated to allow for providing a per-sample scale factor (#799)

* hifiasm copied from fastqc

* hifiasm tests init from fastqc

* meta.yml init; test.yml and main.nf for printing version

* Add hifiasm version printing

* Removed spaced on an empty line

* Reverted hifiasm from main

* Added genomecov scale module

* Updated tagging

* Removed extra module - began merging

* Removed extra module tests

* Updated genomecov to take a scale value

* Updated line endings

* Removed redundant test

* Update tests/modules/bedtools/genomecov/main.nf

Co-authored-by: Harshil Patel <drpatelh@users.noreply.github.com>

* Added checking for existing -bg arg

* Update modules/bedtools/genomecov/main.nf

Co-authored-by: Harshil Patel <drpatelh@users.noreply.github.com>

* Update modules/bedtools/genomecov/main.nf

Co-authored-by: Harshil Patel <drpatelh@users.noreply.github.com>

Co-authored-by: Sviatoslav Sidorov <sviatoslav.sidorov@crick.ac.uk>
Co-authored-by: Svyatoslav Sidorov <svet.sidorov@gmail.com>
Co-authored-by: Harshil Patel <drpatelh@users.noreply.github.com>

modules/bedtools/genomecov/main.nf

@@ -19,7 +19,7 @@ process BEDTOOLS_GENOMECOV {
     }
 
     input:
-    tuple val(meta), path(intervals)
+    tuple val(meta), path(intervals), val(scale)
     path  sizes
     val   extension
 
@@ -29,12 +29,20 @@ process BEDTOOLS_GENOMECOV {
 
     script:
     def prefix     = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+    def args_token = options.args.tokenize()
+    def args       = options.args
+    args += (scale > 0 && scale != 1) ? " -scale $scale" : ""
+
+    if (!args_token.contains('-bg') && (scale > 0 && scale != 1)) {
+        args += " -bg"
+    }
+
     if (intervals.name =~ /\.bam/) {
         """
         bedtools \\
             genomecov \\
             -ibam $intervals \\
-            $options.args \\
+            $args \\
             > ${prefix}.${extension}
 
         cat <<-END_VERSIONS > versions.yml
@@ -48,7 +56,7 @@ process BEDTOOLS_GENOMECOV {
             genomecov \\
             -i $intervals \\
             -g $sizes \\
-            $options.args \\
+            $args \\
             > ${prefix}.${extension}
 
         cat <<-END_VERSIONS > versions.yml

modules/bedtools/genomecov/meta.yml

@@ -20,6 +20,9 @@ input:
         type: file
         description: BAM/BED/GFF/VCF
         pattern: "*.{bam|bed|gff|vcf}"
+    - scale:
+        type: value
+        description: Number containing the scale factor for the output. Set to 1 to disable. Setting to a value other than 1 will also get the -bg bedgraph output format as this is required for this command switch
     - sizes:
         type: file
         description: Tab-delimited table of chromosome names in the first column and chromosome sizes in the second column
@@ -45,3 +48,4 @@ authors:
     - "@sruthipsuresh"
     - "@drpatelh"
     - "@sidorov-si"
+    - "@chris-cheshire"

tests/modules/bedtools/genomecov/main.nf

@@ -4,10 +4,37 @@ nextflow.enable.dsl = 2
 
 include { BEDTOOLS_GENOMECOV } from '../../../../modules/bedtools/genomecov/main.nf' addParams( options: [suffix: '_out'] )
 
-workflow test_bedtools_genomecov {
+workflow test_bedtools_genomecov_noscale {
     input = [
         [ id:'test'],
-        file(params.test_data['sarscov2']['illumina']['test_paired_end_bam'], checkIfExists: true)
+        file(params.test_data['sarscov2']['illumina']['test_paired_end_bam'], checkIfExists: true),
+        1
+    ]
+
+    sizes = []
+    extension = 'txt'
+
+    BEDTOOLS_GENOMECOV ( input, sizes, extension )
+}
+
+workflow test_bedtools_genomecov_nonbam_noscale {
+    input = [
+        [ id:'test'],
+        file(params.test_data['sarscov2']['genome']['baits_bed'], checkIfExists: true),
+        1
+    ]
+
+    sizes = file(params.test_data['sarscov2']['genome']['genome_sizes'], checkIfExists: true)
+    extension = 'txt'
+
+    BEDTOOLS_GENOMECOV ( input, sizes, extension )
+}
+
+workflow test_bedtools_genomecov_scale {
+    input = [
+        [ id:'test'],
+        file(params.test_data['sarscov2']['illumina']['test_paired_end_bam'], checkIfExists: true),
+        0.5
     ]
 
     sizes = file('dummy_chromosome_sizes')
@@ -16,10 +43,11 @@ workflow test_bedtools_genomecov {
     BEDTOOLS_GENOMECOV ( input, sizes, extension )
 }
 
-workflow test_bedtools_genomecov_nonbam {
+workflow test_bedtools_genomecov_nonbam_scale {
     input = [
         [ id:'test'],
-        file(params.test_data['sarscov2']['genome']['baits_bed'], checkIfExists: true)
+        file(params.test_data['sarscov2']['genome']['baits_bed'], checkIfExists: true),
+        0.5
     ]
 
     sizes = file(params.test_data['sarscov2']['genome']['genome_sizes'], checkIfExists: true)

tests/modules/bedtools/genomecov/test.yml

@@ -1,5 +1,5 @@
-- name: bedtools genomecov test_bedtools_genomecov
-  command: nextflow run ./tests/modules/bedtools/genomecov -entry test_bedtools_genomecov -c tests/config/nextflow.config
+- name: bedtools genomecov test_bedtools_genomecov_noscale
+  command: nextflow run ./tests/modules/bedtools/genomecov -entry test_bedtools_genomecov_noscale -c tests/config/nextflow.config
   tags:
     - bedtools
     - bedtools/genomecov
@@ -7,11 +7,29 @@
     - path: output/bedtools/test_out.txt
       md5sum: 66083198daca6c001d328ba9616e9b53
 
-- name: bedtools genomecov test_bedtools_genomecov_nonbam
-  command: nextflow run tests/modules/bedtools/genomecov -entry test_bedtools_genomecov_nonbam -c tests/config/nextflow.config
+- name: bedtools genomecov test_bedtools_genomecov_nonbam_noscale
+  command: nextflow run tests/modules/bedtools/genomecov -entry test_bedtools_genomecov_nonbam_noscale -c tests/config/nextflow.config
   tags:
     - bedtools
     - bedtools/genomecov
   files:
     - path: output/bedtools/test_out.txt
       md5sum: f47b58840087426e5b643d8dfd155c1f
+
+- name: bedtools genomecov test_bedtools_genomecov_scale
+  command: nextflow run ./tests/modules/bedtools/genomecov -entry test_bedtools_genomecov_scale -c tests/config/nextflow.config
+  tags:
+    - bedtools
+    - bedtools/genomecov
+  files:
+    - path: output/bedtools/test_out.txt
+      md5sum: 01291b6e1beab72e046653e709eb0e10
+
+- name: bedtools genomecov test_bedtools_genomecov_nonbam_scale
+  command: nextflow run tests/modules/bedtools/genomecov -entry test_bedtools_genomecov_nonbam_scale -c tests/config/nextflow.config
+  tags:
+    - bedtools
+    - bedtools/genomecov
+  files:
+    - path: output/bedtools/test_out.txt
+      md5sum: de3c59c0ea123bcdbbad27bc0a0a601e