GTDBTK_CLASSIFYWF database improvement (#8220)

* Add new input channel for premade seqid2map file

* Add extra tests for using custom seqid2map file

* Apply suggestions from code review

* Update modules/nf-core/kraken2/build/tests/main.nf.test

Co-authored-by: Simon Pearce <[email protected]>

* Make database supply to GTDBTK classify WF more robust

* Fix metadata

* Update modules/nf-core/gtdbtk/classifywf/meta.yml

Co-authored-by: Jim Downie <[email protected]>

---------

Co-authored-by: Simon Pearce <[email protected]>
Co-authored-by: Jim Downie <[email protected]>

Author: 3 people

modules/nf-core/gtdbtk/classifywf/main.nf

@@ -5,21 +5,21 @@ process GTDBTK_CLASSIFYWF {
     container "${workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ? 'https://depot.galaxyproject.org/singularity/gtdbtk:2.4.0--pyhdfd78af_1' : 'biocontainers/gtdbtk:2.4.0--pyhdfd78af_1'}"
 
     input:
-    tuple val(meta)   , path("bins/*")
-    tuple val(db_name), path("database/*")
+    tuple val(meta), path("bins/*")
+    tuple val(db_name), path(db)
     val use_pplacer_scratch_dir
     path mash_db
 
     output:
-    tuple val(meta), path("gtdbtk.${prefix}.*.summary.tsv")        , emit: summary
-    tuple val(meta), path("gtdbtk.${prefix}.*.classify.tree.gz")   , emit: tree    , optional: true
-    tuple val(meta), path("gtdbtk.${prefix}.*.markers_summary.tsv"), emit: markers , optional: true
-    tuple val(meta), path("gtdbtk.${prefix}.*.msa.fasta.gz")       , emit: msa     , optional: true
-    tuple val(meta), path("gtdbtk.${prefix}.*.user_msa.fasta.gz")  , emit: user_msa, optional: true
-    tuple val(meta), path("gtdbtk.${prefix}.*.filtered.tsv")       , emit: filtered, optional: true
-    tuple val(meta), path("gtdbtk.${prefix}.failed_genomes.tsv")   , emit: failed  , optional: true
-    tuple val(meta), path("gtdbtk.${prefix}.log")                  , emit: log
-    tuple val(meta), path("gtdbtk.${prefix}.warnings.log")         , emit: warnings
+    tuple val(meta), path("gtdbtk.${prefix}.*.summary.tsv"), emit: summary
+    tuple val(meta), path("gtdbtk.${prefix}.*.classify.tree.gz"), emit: tree, optional: true
+    tuple val(meta), path("gtdbtk.${prefix}.*.markers_summary.tsv"), emit: markers, optional: true
+    tuple val(meta), path("gtdbtk.${prefix}.*.msa.fasta.gz"), emit: msa, optional: true
+    tuple val(meta), path("gtdbtk.${prefix}.*.user_msa.fasta.gz"), emit: user_msa, optional: true
+    tuple val(meta), path("gtdbtk.${prefix}.*.filtered.tsv"), emit: filtered, optional: true
+    tuple val(meta), path("gtdbtk.${prefix}.failed_genomes.tsv"), emit: failed, optional: true
+    tuple val(meta), path("gtdbtk.${prefix}.log"), emit: log
+    tuple val(meta), path("gtdbtk.${prefix}.warnings.log"), emit: warnings
     path ("versions.yml"), emit: versions
 
     when:
@@ -28,11 +28,12 @@ process GTDBTK_CLASSIFYWF {
     script:
     def args = task.ext.args ?: ''
     def pplacer_scratch = use_pplacer_scratch_dir ? "--scratch_dir pplacer_tmp" : ""
-    def mash_mode       = mash_db                 ? "--mash_db ${mash_db}"      : "--skip_ani_screen"
+    def mash_mode = mash_db ? "--mash_db ${mash_db}" : "--skip_ani_screen"
     prefix = task.ext.prefix ?: "${meta.id}"
 
     """
-    export GTDBTK_DATA_PATH="\${PWD}/database"
+    export GTDBTK_DATA_PATH="\$(find -L ${db} -name 'metadata' -type d -exec dirname {} \\;)"#
+
     if [ ${pplacer_scratch} != "" ] ; then
         mkdir pplacer_tmp
     fi

modules/nf-core/gtdbtk/classifywf/meta.yml

@@ -27,13 +27,18 @@ input:
           e.g. [ id:'test', single_end:false,  assembler:'spades' ]
     - '"bins/*"':
         type: file
-        description: bins
+        description: A list of one or more bins in FASTA format for classification
+        pattern: "*.{fasta,fna,fas,fa}{,.gz}"
   - - db_name:
         type: string
         description: The name of the GTDB database to use.
-    - '"database/*"':
+    - db:
         type: file
-        description: GTDB database
+        description: |
+          Path to a directory containing a GDTB database, as uncompressed from from the 'full package' gtdbdtk_data.tar.gz file.
+          You can give the 'release<version number>' directory here.
+          Must contain the 'metadata' subdirectory
+        pattern: "release[0-9]+\/"
   - - use_pplacer_scratch_dir:
         type: boolean
         description: Set to true to reduce pplacer memory usage by writing to disk (slower)