Dastool update to allow non-gzipped inputs (#1458)

* fix: remove left-over unnecessary code

* Make gzipping optional for DAS_Tool scaffolds2bin

* Add optional unzipping

* Make gunzip optional for DAS_Tool scaffolds2bin

* Apply suggestions from code review

Co-authored-by: Daniel Straub <42973691+d4straub@users.noreply.github.com>

* Update modules/dastool/scaffolds2bin/meta.yml

Co-authored-by: Daniel Straub <42973691+d4straub@users.noreply.github.com>

modules/dastool/scaffolds2bin/main.nf

@@ -22,9 +22,10 @@ process DASTOOL_SCAFFOLDS2BIN {
     def args = task.ext.args ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
     def file_extension = extension ? extension : "fasta"
-
+    def clean_fasta = fasta.toString() - ".gz"
+    def decompress_fasta = fasta.toString() == clean_fasta ? "" : "gunzip -q -f $fasta"
     """
-    gunzip -f *.${file_extension}.gz
+    $decompress_fasta
 
     Fasta_to_Scaffolds2Bin.sh \\
         $args \\

modules/dastool/scaffolds2bin/meta.yml

@@ -30,14 +30,11 @@ input:
         e.g. [ id:'test', single_end:false ]
   - fasta:
       type: file
-      description: Fasta of list of fasta files recommended to be gathered via with .collect() of bins
+      description: Fasta or list of fasta files recommended to be gathered via with .collect() of bins
-      pattern: "*.{fa,fas,fasta}"
+      pattern: "*.{fa,fa.gz,fas,fas.gz,fna,fna.gz,fasta,fasta.gz}"
-  - binner:
-      type: val
-      description: Name of the binning software (optional)
   - extension:
       type: val
-      description: Fasta file extension (fa | fas | fasta | ...)
+      description: Fasta file extension (fa | fas | fasta | ...), but without .gz suffix, even if gzipped input.
 
 output:
   - meta:

tests/modules/dastool/scaffolds2bin/main.nf

@@ -2,6 +2,7 @@
 
 nextflow.enable.dsl = 2
 
+include { GUNZIP                                } from '../../../../modules/gunzip/main.nf'
 include { METABAT2_METABAT2                     } from '../../../../modules/metabat2/metabat2/main.nf'
 include { METABAT2_JGISUMMARIZEBAMCONTIGDEPTHS  } from '../../../../modules/metabat2/jgisummarizebamcontigdepths/main.nf'
 include { DASTOOL_SCAFFOLDS2BIN                 } from '../../../../modules/dastool/scaffolds2bin/main.nf'
@@ -23,3 +24,25 @@ workflow test_dastool_scaffolds2bin {
 
     DASTOOL_SCAFFOLDS2BIN ( METABAT2_METABAT2.out.fasta.collect(), "fa")
 }
+
+workflow test_dastool_scaffolds2bin_ungzipped {
+
+    input_depth = [ [ id:'test', single_end:false ], // meta map
+              file(params.test_data['bacteroides_fragilis']['illumina']['test1_paired_end_sorted_bam'], checkIfExists: true),
+              file(params.test_data['bacteroides_fragilis']['illumina']['test1_paired_end_sorted_bam_bai'], checkIfExists: true) ]
+
+
+    METABAT2_JGISUMMARIZEBAMCONTIGDEPTHS ( input_depth )
+
+    Channel.fromPath(params.test_data['bacteroides_fragilis']['genome']['genome_fna_gz'], checkIfExists: true)
+        .map { it -> [[ id:'test', single_end:false ], it] }
+        .join(METABAT2_JGISUMMARIZEBAMCONTIGDEPTHS.out.depth)
+        .set { input_metabat2 }
+
+    METABAT2_METABAT2 ( input_metabat2 )
+
+    // TODO test unzipped input files
+    ch_input_2_scaffolds2bin = GUNZIP( METABAT2_METABAT2.out.fasta  ).gunzip
+
+    DASTOOL_SCAFFOLDS2BIN ( ch_input_2_scaffolds2bin, "fa")
+}

tests/modules/dastool/scaffolds2bin/test.yml

@@ -1,14 +1,20 @@
 - name: dastool scaffolds2bin test_dastool_scaffolds2bin
-  command: nextflow run ./tests/modules/dastool/scaffolds2bin -entry test_dastool_scaffolds2bin -c ./tests/config/nextflow.config -c ./tests/modules/dastool/scaffolds2bin/nextflow.config
+  command: nextflow run tests/modules/dastool/scaffolds2bin -entry test_dastool_scaffolds2bin -c tests/config/nextflow.config
   tags:
     - dastool
     - dastool/scaffolds2bin
   files:
     - path: output/dastool/test.tsv
       md5sum: 6e46c0be14dded7cb13af38f54feea47
-    - path: output/metabat2/bins/test.1.fa.gz
+    - path: output/dastool/versions.yml
-      md5sum: 2b297bf557cc3831b800348859331268
+      md5sum: d0831ed159eb5a1a1565d1d211012ad6
-    - path: output/metabat2/test.tsv.gz
+- name: dastool scaffolds2bin test_dastool_scaffolds2bin_ungzipped
-      md5sum: 619338fa5019e361d5545ce385a6961f
+  command: nextflow run tests/modules/dastool/scaffolds2bin -entry test_dastool_scaffolds2bin_ungzipped -c tests/config/nextflow.config
-    - path: output/metabat2/test.txt.gz
+  tags:
-      md5sum: 745a0446af6ef68b930975e9ce5a95d6
+    - dastool
+    - dastool/scaffolds2bin
+  files:
+    - path: output/dastool/test.tsv
+      md5sum: 6e46c0be14dded7cb13af38f54feea47
+    - path: output/dastool/versions.yml
+      md5sum: da58e477b7f4c16a9ea495ec1a4a4d4f