Update centrifuge/centrifuge (#1495)

modules/centrifuge/centrifuge/main.nf

@@ -1,4 +1,4 @@
-process CENTRIFUGE {
+process CENTRIFUGE_CENTRIFUGE {
     tag "$meta.id"
     label 'process_high'
 
@@ -10,7 +10,6 @@ process CENTRIFUGE {
     input:
     tuple val(meta), path(reads)
     path db
-    val db_name
     val save_unaligned
     val save_aligned
     val sam_format
@@ -18,7 +17,6 @@ process CENTRIFUGE {
     output:
     tuple val(meta), path('*report.txt')                 , emit: report
     tuple val(meta), path('*results.txt')                , emit: results
-    tuple val(meta), path('*kreport.txt')                , emit: kreport
     tuple val(meta), path('*.sam')                       , optional: true, emit: sam
     tuple val(meta), path('*.mapped.fastq{,.1,.2}.gz')   , optional: true, emit: fastq_mapped
     tuple val(meta), path('*.unmapped.fastq{,.1,.2}.gz') , optional: true, emit: fastq_unmapped
@@ -31,7 +29,6 @@ process CENTRIFUGE {
     def args = task.ext.args ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
     def paired = meta.single_end ? "-U ${reads}" :  "-1 ${reads[0]} -2 ${reads[1]}"
-    def db_name = db.toString().replace(".tar.gz","")
     def unaligned = ''
     def aligned = ''
     if (meta.single_end) {
@@ -43,8 +40,10 @@ process CENTRIFUGE {
     }
     def sam_output = sam_format ? "--out-fmt 'sam'" : ''
     """
+    ## we add "-no-name ._" to ensure silly Mac OSX metafiles files aren't included
+    db_name=`find -L ${db} -name "*.1.cf" -not -name "._*"  | sed 's/.1.cf//'`
     centrifuge \\
-        -x ${db}/${db_name} \\
+        -x \$db_name \\
         -p $task.cpus \\
         $paired \\
         --report-file ${prefix}.report.txt \\
@@ -53,7 +52,6 @@ process CENTRIFUGE {
         $aligned \\
         $sam_output \\
         $args
-    centrifuge-kreport -x $db_name ${prefix}.results.txt > ${prefix}.kreport.txt
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":

modules/centrifuge/centrifuge/meta.yml

@@ -1,4 +1,4 @@
-name: centrifuge
+name: centrifuge_centrifuge
 description: Classifies metagenomic sequence data
 keywords:
   - classify
@@ -25,11 +25,7 @@ input:
         respectively.
   - db:
       type: directory
-      description: Centrifuge database in .tar.gz format
-      pattern: "*.tar.gz"
-  - db_name:
-      type: string
-      description: Centrifuge database filenames without the suffix ".cf"
+      description: Path to directory containing centrifuge database files
   - save_unaligned:
       type: value
       description: If true unmapped fastq files are saved
@@ -52,12 +48,6 @@ output:
       description: |
         File containing classification results
       pattern: "*.{results.txt}"
-  - kreport:
-      type: file
-      description: |
-        File containing kraken-style report from centrifuge
-        out files.
-      pattern: "*.{kreport.txt}"
   - fastq_unmapped:
       type: file
       description: Unmapped fastq files

tests/config/pytest_modules.yml

@@ -395,9 +395,9 @@ cellranger/mkref:
   - modules/cellranger/gtf/**
   - tests/modules/cellranger/gtf/**
 
-centrifuge:
-  - modules/centrifuge/**
-  - tests/modules/centrifuge/**
+centrifuge/centrifuge:
+  - modules/centrifuge/centrifuge/**
+  - tests/modules/centrifuge/centrifuge/**
 
 checkm/lineagewf:
   - modules/checkm/lineagewf/**

tests/modules/centrifuge/centrifuge/main.nf

@@ -2,37 +2,36 @@
 
 nextflow.enable.dsl = 2
 
-include { UNTAR      } from '../../../modules/untar/main.nf'
-include { CENTRIFUGE } from '../../../modules/centrifuge/main.nf'
+include { UNTAR                 } from '../../../../modules/untar/main.nf'
+include { CENTRIFUGE_CENTRIFUGE } from '../../../../modules/centrifuge/centrifuge/main.nf'
 
-workflow test_centrifuge_single_end {
+workflow test_centrifuge_centrifuge_single_end {
     input = [ [ id:'test', single_end:true ], // meta map
               [ file(params.test_data['sarscov2']['illumina']['test_1_fastq_gz'], checkIfExists: true) ]
             ]
     db    =  [ [], file('https://raw.githubusercontent.com/nf-core/test-datasets/modules/data/delete_me/minigut_cf.tar.gz', checkIfExists: true) ]
-    db_name = "minigut_cf"
     save_unaligned = true
     save_aligned = false
     sam_format = false
 
     UNTAR ( db )
-    CENTRIFUGE ( input, UNTAR.out.untar.map{ it[1] },db_name, save_unaligned, save_aligned, sam_format )
+    CENTRIFUGE_CENTRIFUGE ( input, UNTAR.out.untar.map{ it[1] }, save_unaligned, save_aligned, sam_format )
 
 }
 
-workflow test_centrifuge_paired_end {
+workflow test_centrifuge_centrifuge_paired_end {
     input = [ [ id:'test', single_end:false ], // meta map
               [ file(params.test_data['sarscov2']['illumina']['test_1_fastq_gz'], checkIfExists: true),
                 file(params.test_data['sarscov2']['illumina']['test_2_fastq_gz'], checkIfExists: true) ]
             ]
      db    =  [ [], file('https://raw.githubusercontent.com/nf-core/test-datasets/modules/data/delete_me/minigut_cf.tar.gz', checkIfExists: true) ]
-     db_name = "minigut_cf"
+     //db_name = "minigut_cf"
      save_unaligned = true
      save_aligned = false
      sam_format = false
 
     UNTAR ( db )
-    CENTRIFUGE ( input, UNTAR.out.untar.map{ it[1] }, db_name, save_unaligned, save_aligned, sam_format )
+    CENTRIFUGE_CENTRIFUGE ( input, UNTAR.out.untar.map{ it[1] }, save_unaligned, save_aligned, sam_format )
 
 
 }

tests/modules/centrifuge/centrifuge/test.yml

@@ -1,20 +1,20 @@
-- name: centrifuge test_centrifuge_single_end
-  command: nextflow run tests/modules/centrifuge -entry test_centrifuge_single_end -c tests/config/nextflow.config
+- name: centrifuge centrifuge test_centrifuge_centrifuge_single_end
+  command: nextflow run tests/modules/centrifuge/centrifuge -entry test_centrifuge_centrifuge_single_end -c tests/config/nextflow.config
   tags:
     - centrifuge
+    - centrifuge/centrifuge
   files:
-    - path: output/centrifuge/test.kreport.txt
     - path: output/centrifuge/test.report.txt
     - path: output/centrifuge/test.results.txt
     - path: output/centrifuge/test.unmapped.fastq.gz
     - path: output/centrifuge/versions.yml
 
-- name: centrifuge test_centrifuge_paired_end
-  command: nextflow run tests/modules/centrifuge -entry test_centrifuge_paired_end -c tests/config/nextflow.config
+- name: centrifuge centrifuge test_centrifuge_centrifuge_paired_end
+  command: nextflow run tests/modules/centrifuge/centrifuge -entry test_centrifuge_centrifuge_paired_end -c tests/config/nextflow.config
   tags:
     - centrifuge
+    - centrifuge/centrifuge
   files:
-    - path: output/centrifuge/test.kreport.txt
     - path: output/centrifuge/test.report.txt
     - path: output/centrifuge/test.results.txt
     - path: output/centrifuge/test.unmapped.fastq.1.gz