Update MALT to older version due to major bug in most recent version (#1796)

* Rollback version

* Start work rollback to 0.4.1

* Further changes

* Re do dynamic flag creation

* Prettier.

modules/malt/build/main.nf

@@ -2,16 +2,17 @@ process MALT_BUILD {
 
     label 'process_high'
 
-    conda (params.enable_conda ? "bioconda::malt=0.53" : null)
+    conda (params.enable_conda ? "bioconda::malt=0.41" : null)
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/malt:0.53--hdfd78af_0' :
+        'https://depot.galaxyproject.org/singularity/malt:0.41--1' :
-        'quay.io/biocontainers/malt:0.53--hdfd78af_0' }"
+        'quay.io/biocontainers/malt:0.41--1' }"
 
     input:
     path fastas
     val seq_type
-    path gff
+    path mapping_file
-    path map_db
+    val mapping_type
+    val mapping_db
 
     output:
     path "malt_index/"   , emit: index
@@ -29,19 +30,36 @@ process MALT_BUILD {
     } else {
         avail_mem = task.memory.giga
     }
-    def igff = gff ? "-igff ${gff}" : ""
+
+    def valid_db = ['eggnog', 'interpro2go', 'kegg', 'seed', 'taxonomy']
+    if ( !valid_db.contains(mapping_db) )  { error "Unrecognised mapping database value for MALT_BUILD. Options: eggnog, interpro2go, kegg, seed, taxonomy" }
+
+    switch ( "${mapping_type}" ) {
+        case "gi":
+        mapping_prefix = "-g"; break
+        case "ref":
+        if ( mapping_db == "taxonomy" ) {
+            mapping_prefix = '-a'
+        } else {
+            mapping_prefix = "-r"
+        };break
+        case "syn":
+        mapping_prefix = "-s"; break
+        default:
+        error '[MALT_BUILD] Mapping type not recognised. Options: gi, ref, syn'; break
+    }
+
+    type_flag = mapping_prefix + '2' + mapping_db + " " + mapping_file
 
     """
     malt-build \\
-        -J-Xmx${avail_mem}g \\
         -v \\
         --input ${fastas.join(' ')} \\
         -s $seq_type \\
-        $igff \\
         -d 'malt_index/' \\
         -t $task.cpus \\
         $args \\
-        -mdb ${map_db}/*.db |&tee malt-build.log
+        $type_flag |&tee malt-build.log
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":

modules/malt/build/meta.yml

@@ -22,20 +22,24 @@ tools:
 input:
   - fastas:
       type: file
-      description: Directory of, or FASTA reference files for indexing
+      description: Directory of, or list of FASTA reference files for indexing
       pattern: "*/|*.fasta"
   - seq_type:
       type: string
       description: Type of input data
       pattern: "DNA|Protein"
-  - gff:
+  - mapping_file:
       type: file
-      description: Directory of, or GFF3 files of input FASTA files
+      description: An unzipped MEGAN mapping file in .abin format (pre-2020), downloadable from https://software-ab.informatik.uni-tuebingen.de/download/megan6/old.html
-      pattern: "*/|*.gff|*.gff3"
+      pattern: "*.abin"
-  - map_db:
+  - mapping_type:
-      type: file
+      type: value
-      description: MEGAN .db file from https://software-ab.informatik.uni-tuebingen.de/download/megan6/welcome.html
+      description: What type of accession to use for database construction.
-      pattern:
+      pattern: "gi|ref|syn"
+  - mapping_type:
+      type: value
+      description: Which database the mapping file is derived from.
+      pattern: "eggnog|interpro2go|kegg|seed|taxonomy"
 
 output:
   - versions:

modules/malt/run/main.nf

@@ -2,10 +2,10 @@ process MALT_RUN {
     tag "$meta.id"
     label 'process_high'
 
-    conda (params.enable_conda ? "bioconda::malt=0.53" : null)
+    conda (params.enable_conda ? "bioconda::malt=0.41" : null)
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/malt:0.53--hdfd78af_0' :
+        'https://depot.galaxyproject.org/singularity/malt:0.41--1' :
-        'quay.io/biocontainers/malt:0.53--hdfd78af_0' }"
+        'quay.io/biocontainers/malt:0.41--1' }"
 
     input:
     tuple val(meta), path(fastqs)
@@ -33,7 +33,6 @@ process MALT_RUN {
 
     """
     malt-run \\
-        -J-Xmx${avail_mem}g \\
         -t $task.cpus \\
         -v \\
         -o . \\

tests/modules/malt/build_test/main.nf

@@ -6,21 +6,12 @@ include { UNZIP } from '../../../../modules/unzip/main.nf'
 include { MALT_BUILD } from '../../../../modules/malt/build/main.nf'
 
 workflow test_malt_build {
-    fastas = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+    fastas        = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
-    seq_type = "DNA"
+    seq_type      = "DNA"
-    gff = []
+    map_accession = [ [], file("https://software-ab.informatik.uni-tuebingen.de/download/megan6/nucl_acc2tax-Jul2019.abin.zip", checkIfExists: true) ]
-    map_db = [ [], file("https://software-ab.informatik.uni-tuebingen.de/download/megan6/megan-nucl-Jan2021.db.zip", checkIfExists: true) ]
+    mapping_type  = 'ref'
+    mapping_db    = 'taxonomy'
 
-    UNZIP ( map_db )
+    UNZIP ( map_accession )
-    MALT_BUILD ( fastas, seq_type, gff, UNZIP.out.unzipped_archive.map{ it[1] } )
+    MALT_BUILD ( fastas, seq_type, UNZIP.out.unzipped_archive.map{ it[1] }, "ref", "taxonomy" )
-}
-
-workflow test_malt_build_gff {
-    fastas = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
-    seq_type = "DNA"
-    gff = file(params.test_data['sarscov2']['genome']['genome_gff3'], checkIfExists: true)
-    map_db = [ [], file("https://software-ab.informatik.uni-tuebingen.de/download/megan6/megan-nucl-Jan2021.db.zip", checkIfExists: true) ]
-
-    UNZIP ( map_db )
-    MALT_BUILD ( fastas, seq_type, gff, UNZIP.out.unzipped_archive.map{ it[1] } )
 }

tests/modules/malt/build_test/test.yml

@@ -1,13 +1,16 @@
-- name: malt build
+- name: malt build test_malt_build
-  command: nextflow run ./tests/modules/malt/build_test -entry test_malt_build -c ./tests/config/nextflow.config -c ./tests/modules/malt/build/nextflow.config
+  command: nextflow run ./tests/modules/malt/build -entry test_malt_build -c ./tests/config/nextflow.config  -c ./tests/modules/malt/build/nextflow.config
   tags:
     - malt
     - malt/build
   files:
+    - path: output/malt/malt-build.log
+      contains:
+        - "Peak memory"
     - path: output/malt/malt_index/index0.idx
       md5sum: 1954f2c00b418d00112829b0a6adb8ce
     - path: output/malt/malt_index/ref.db
-      md5sum: 772a09aeb162515485b037604399f2bd
+      md5sum: 1fb74eccd5400fb23454454da1bd4c0c
     - path: output/malt/malt_index/ref.idx
       md5sum: 7dea362b3fac8e00956a4952a3d4f474
     - path: output/malt/malt_index/ref.inf
@@ -15,34 +18,10 @@
     - path: output/malt/malt_index/table0.db
     - path: output/malt/malt_index/table0.idx
     - path: output/malt/malt_index/taxonomy.idx
-      md5sum: bb335e7c378a5bd85761b6eeed16d984
+      md5sum: 13aa81314892b5537319fca2a63a2c31
     - path: output/malt/malt_index/taxonomy.map
-      md5sum: f6b05bbab2149e388cb769098e14d433
+      md5sum: 1e972302ae6d705b8abb377cfafd380a
     - path: output/malt/malt_index/taxonomy.tre
-      md5sum: bde26a1fff5c63d3046d3863607a1e70
+      md5sum: 79c2322475a8eebc57607ff36c1e4728
-- name: malt build gff
+    - path: output/unzip/nucl_acc2tax-Jul2019.abin/nucl_acc2tax-Jul2019.abin
-  command: nextflow run ./tests/modules/malt/build_test -entry test_malt_build_gff -c ./tests/config/nextflow.config -c ./tests/modules/malt/build/nextflow.config
+      md5sum: b2deb1df50ebcec4de71a6beec6226d7
-  tags:
-    - malt
-    - malt/build
-  files:
-    - path: output/malt/malt_index/aadd.dbx
-      md5sum: 4e2ed57e713d5372bd09350f447cdf53
-    - path: output/malt/malt_index/aadd.idx
-      md5sum: 0994061bc8673ebd283fa6546c3dd12c
-    - path: output/malt/malt_index/index0.idx
-      md5sum: 1954f2c00b418d00112829b0a6adb8ce
-    - path: output/malt/malt_index/ref.db
-      md5sum: 772a09aeb162515485b037604399f2bd
-    - path: output/malt/malt_index/ref.idx
-      md5sum: 7dea362b3fac8e00956a4952a3d4f474
-    - path: output/malt/malt_index/ref.inf
-      md5sum: b146842067cf278ef1d23e6c2e7c0c35
-    - path: output/malt/malt_index/table0.db
-    - path: output/malt/malt_index/table0.idx
-    - path: output/malt/malt_index/taxonomy.idx
-      md5sum: bb335e7c378a5bd85761b6eeed16d984
-    - path: output/malt/malt_index/taxonomy.map
-      md5sum: f6b05bbab2149e388cb769098e14d433
-    - path: output/malt/malt_index/taxonomy.tre
-      md5sum: bde26a1fff5c63d3046d3863607a1e70

tests/modules/malt/run/main.nf

@@ -8,18 +8,19 @@ include { MALT_RUN   } from '../../../../modules/malt/run/main.nf'
 
 workflow test_malt_run {
 
-    fastas = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+    fastas        = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
-    gff = file(params.test_data['sarscov2']['genome']['genome_gff3'], checkIfExists: true)
+    seq_type      = "DNA"
-    seq_type = "DNA"
+    map_accession = [ [], file("https://software-ab.informatik.uni-tuebingen.de/download/megan6/nucl_acc2tax-Jul2019.abin.zip", checkIfExists: true) ]
-    map_db = [ [], file("https://software-ab.informatik.uni-tuebingen.de/download/megan6/megan-nucl-Jan2021.db.zip", checkIfExists: true) ]
+    mapping_type  = 'ref'
+    mapping_db    = 'taxonomy'
     input = [
         [ id:'test', single_end:false ], // meta map
         file(params.test_data['sarscov2']['illumina']['test_1_fastq_gz'], checkIfExists: true)
     ]
     mode = "BlastN"
 
-    UNZIP ( map_db )
+    UNZIP ( map_accession )
-    MALT_BUILD ( fastas, seq_type, gff, UNZIP.out.unzipped_archive.map { it[1] } )
+    MALT_BUILD ( fastas, seq_type, UNZIP.out.unzipped_archive.map{ it[1] }, "ref", "taxonomy" )
     MALT_RUN ( input, mode, MALT_BUILD.out.index )
 }
 

tests/modules/malt/run/test.yml

@@ -1,8 +1,10 @@
-- name: malt run
+- name: malt run test_malt_run
-  command: nextflow run ./tests/modules/malt/run -entry test_malt_run -c ./tests/config/nextflow.config -c ./tests/modules/malt/run/nextflow.config
+  command: nextflow run ./tests/modules/malt/run -entry test_malt_run -c ./tests/config/nextflow.config  -c ./tests/modules/malt/run/nextflow.config
   tags:
-    - malt
     - malt/run
+    - malt
   files:
-    - path: output/malt/test_1.rma6
     - path: output/malt/test-malt-run.log
+      contains:
+        - "Peak memory"
+    - path: output/malt/test_1.rma6