new logic

modules/local/check_duplicates.nf

@@ -1,19 +1,23 @@
 process CHECK_DUPLICATES {
     label 'process_low'
+    tag "${meta.id}"
 
     conda "conda-forge::pigz=2.6"
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
         'https://depot.galaxyproject.org/singularity/ubuntu:24.04' :
         'biocontainers/ubuntu:24.04' }"
 
     input:
-    path fnas
+    tuple val(meta), path(fna)
 
     output:
     path "duplicates.txt"            , emit: duplicates_file, optional: true
+    tuple val(meta), path(fna)       , emit: fna, optional: true
     path "versions.yml"              , emit: versions
 
     script:
+    prefix = task.ext.prefix ?: meta.id
+
     """
     # Find duplicate contig names across all files
     zgrep -H '>' *.fna.gz | sed 's/^[^:]*://' | sort | uniq -d > duplicate_contig_names.txt

nextflow.config

@@ -18,6 +18,7 @@ params {
     checkm_metadata             = null
     gtdbtk_metadata             = null
     se_reads                    = false
+    check_duplicates            = true
     rename_contigs              = false
 
     // Trimming

nextflow_schema.json

@@ -62,6 +62,11 @@
                     "description": "Activate when you work with single ends reads.",
                     "fa_icon": "fas fa-arrows-alt-h"
                 },
+                "check_duplicates": {
+                    "type": "boolean",
+                    "default": true,
+                    "description": "Always active. If it finds duplicates in your contigs files, it will return an error with suggestions."
+                },
                 "rename_contigs": {
                     "type": "boolean",
                     "description": "Activate when you want rename to contig names. Useful when you run magmap after nf-core/mag or in any other case when there's a risk you have the same name of contigs in different genomes."

workflows/magmap.nf

@@ -62,10 +62,23 @@ workflow MAGMAP {
     //
     // Check presence of duplicates contigs in the local genome collection
     //
-    CHECK_DUPLICATES(ch_genomeinfo.map{ it.genome_fna }.collect())
-    ch_versions = ch_versions.mix(CHECK_DUPLICATES.out.versions)
+    if (params.check_duplicates) {
+        CHECK_DUPLICATES(ch_genomeinfo.map{ it.genome_fna }.collect().map { [ [id: 'test'], it ] } )
+        ch_versions = ch_versions.mix(CHECK_DUPLICATES.out.versions)
+
+        // Check for duplicates and emit a value to rename_trigger if needed
+        CHECK_DUPLICATES.out.duplicates_file
+        .countLines()
+        .map { it.toInteger() }
+        .subscribe { count ->
+            if (count > 0) {
+                error """Your genomes have duplicate contig names. Either set --rename_duplicates and resume the pipeline to let the pipeline fix this, or fix the genome files manually and resume the pipeline. NB! deactivate params.check_duplicates and resume the pipeline."""
+            }
+        }
+    }
 
-    if( params.rename_contigs ) {
+    // RENAME_CONTIGS(rename_input)
+    if(  params.rename_contigs ) {
         RENAME_CONTIGS( ch_genomeinfo.map{ [ it.accno, it.genome_fna ] } )
         ch_versions = ch_versions.mix(RENAME_CONTIGS.out.versions)