Upgraded getchromsizes to support meta and output the gzi index (#2046)

* This process is single-threaded

* Added meta to match the other nf-core modules

* "custom" is not a great name

* Newer modules have a stub

* To match the other modules

* self promotion

* Output the gzi file too, if there is one

* More Fasta extensions

Co-authored-by: James A. Fellows Yates <jfy133@gmail.com>

Co-authored-by: James A. Fellows Yates <jfy133@gmail.com>

modules/custom/getchromsizes/main.nf

@@ -1,6 +1,6 @@
 process CUSTOM_GETCHROMSIZES {
     tag "$fasta"
-    label 'process_low'
+    label 'process_single'
 
     conda (params.enable_conda ? "bioconda::samtools=1.15.1" : null)
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
@@ -8,12 +8,13 @@ process CUSTOM_GETCHROMSIZES {
         'quay.io/biocontainers/samtools:1.15.1--h1170115_0' }"
 
     input:
-    path fasta
+    tuple val(meta), path(fasta)
 
     output:
-    path '*.sizes'      , emit: sizes
+    tuple val(meta), path ("*.sizes"), emit: sizes
-    path '*.fai'        , emit: fai
+    tuple val(meta), path ("*.fai")  , emit: fai
-    path  "versions.yml", emit: versions
+    tuple val(meta), path ("*.gzi")  , emit: gzi, optional: true
+    path  "versions.yml"             , emit: versions
 
     when:
     task.ext.when == null || task.ext.when
@@ -26,7 +27,18 @@ process CUSTOM_GETCHROMSIZES {
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
-        custom: \$(echo \$(samtools --version 2>&1) | sed 's/^.*samtools //; s/Using.*\$//')
+        getchromsizes: \$(echo \$(samtools --version 2>&1) | sed 's/^.*samtools //; s/Using.*\$//')
+    END_VERSIONS
+    """
+
+    stub:
+    """
+    touch ${fasta}.fai
+    touch ${fasta}.sizes
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        getchromsizes: \$(echo \$(samtools --version 2>&1) | sed 's/^.*samtools //; s/Using.*\$//')
     END_VERSIONS
     """
 }

modules/custom/getchromsizes/meta.yml

@@ -14,12 +14,22 @@ tools:
       licence: ["MIT"]
 
 input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
   - fasta:
       type: file
       description: FASTA file
-      pattern: "*.{fasta}"
+      pattern: "*.{fa,fasta,fna,fas}"
 
 output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
   - sizes:
       type: file
       description: File containing chromosome lengths
@@ -28,11 +38,16 @@ output:
       type: file
       description: FASTA index file
       pattern: "*.{fai}"
+  - gzi:
+      type: file
+      description: Optional gzip index file for compressed inputs
+      pattern: "*.gzi"
   - versions:
       type: file
-      description: File containing software version
+      description: File containing software versions
       pattern: "versions.yml"
 
 authors:
   - "@tamara-hodgetts"
   - "@chris-cheshire"
+  - "@muffato"

tests/modules/custom/getchromsizes/main.nf

@@ -5,8 +5,17 @@ nextflow.enable.dsl = 2
 include { CUSTOM_GETCHROMSIZES } from '../../../../modules/custom/getchromsizes/main.nf'
 
 workflow test_custom_getchromsizes {
-    
+
-    input = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true) 
+    input = [ [ id:'test', single_end:false ], // meta map
+              file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true) ]
+
+    CUSTOM_GETCHROMSIZES ( input )
+}
+
+workflow test_custom_getchromsizes_bgzip {
+
+    input = [ [ id:'test', single_end:false ], // meta map
+              file(params.test_data['sarscov2']['genome']['genome_fasta_gz'], checkIfExists: true) ]
 
     CUSTOM_GETCHROMSIZES ( input )
 }

tests/modules/custom/getchromsizes/test.yml

@@ -8,3 +8,19 @@
       md5sum: 9da2a56e2853dc8c0b86a9e7229c9fe5
     - path: output/custom/genome.fasta.sizes
       md5sum: a57c401f27ae5133823fb09fb21c8a3c
+    - path: output/custom/versions.yml
+      md5sum: 3e4a23a0852f4ec34296224d87446d9a
+- name: custom getchromsizes_bgzip
+  command: nextflow run ./tests/modules/custom/getchromsizes -entry test_custom_getchromsizes_bgzip -c ./tests/config/nextflow.config -c ./tests/modules/custom/getchromsizes/nextflow.config
+  tags:
+    - custom
+    - custom/getchromsizes
+  files:
+    - path: output/custom/genome.fasta.gz.fai
+      md5sum: 9da2a56e2853dc8c0b86a9e7229c9fe5
+    - path: output/custom/genome.fasta.gz.gzi
+      md5sum: 7dea362b3fac8e00956a4952a3d4f474
+    - path: output/custom/genome.fasta.gz.sizes
+      md5sum: a57c401f27ae5133823fb09fb21c8a3c
+    - path: output/custom/versions.yml
+      md5sum: 22871934dfac30a6109068fd79b2d0ba