Check broken links

modules/local/check_broken_links.nf

@@ -0,0 +1,32 @@
+process CHECK_BROKEN_LINKS {
+    label 'process_long'
+
+    conda "conda-forge::pigz=2.6"
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/mulled-v2-5799ab18b5fc681e75923b2450abaa969907ec98:941789bd7fe00db16531c26de8bf3c5c985242a5-0' :
+        'biocontainers/mulled-v2-5799ab18b5fc681e75923b2450abaa969907ec98:941789bd7fe00db16531c26de8bf3c5c985242a5-0' }"
+
+    input:
+    tuple val(accno), val(genome_fna), val(genome_gff)
+
+    output:
+    tuple val(accno), val(genome_fna), val(genome_gff), emit: valid_genomes_ch
+    path "versions.yml"                               , emit: versions
+
+    script:
+    """
+    # Use curl to check if the URL returns 404
+    if curl -Is "${genome_fna}" | grep -q "404 Not Found"; then
+        echo "Broken link: ${genome_fna}"
+        exit 0  # Exit successfully but don't emit anything
+    else
+        # Correctly echo the variables
+        echo "{accno:'${accno}', genome_fna:'${genome_fna}', genome_gff:'${genome_gff}'}"
+    fi
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        pigz: \$( pigz --version 2>&1 | sed 's/^pigz //' )
+    END_VERSIONS
+    """
+}

subworkflows/local/sourmash.nf

@@ -6,6 +6,7 @@ include { SOURMASH_GATHER                   } from '../../modules/nf-core/sourma
 include { SOURMASH_SKETCH as GENOMES_SKETCH } from '../../modules/nf-core/sourmash/sketch/main'
 include { SOURMASH_INDEX  as GENOMES_INDEX  } from '../../modules/nf-core/sourmash/index/main'
 include { SOURMASH_SKETCH as SAMPLES_SKETCH } from '../../modules/nf-core/sourmash/sketch/main'
+include { CHECK_BROKEN_LINKS                } from '../../modules/local/check_broken_links'
 
 workflow SOURMASH {
     take:
@@ -140,9 +141,25 @@ workflow SOURMASH {
             .mix(ch_matching_user_non_ncbi_genomes)
             .set { ch_filtered_genomes }
 
+    // Check presence of broken links
+    CHECK_BROKEN_LINKS(ch_filtered_genomes)
+    ch_versions = ch_versions.mix(CHECK_BROKEN_LINKS.out.versions)
+
+    // Filter out null entries after link validation
+    ch_valid_genomes_filtered =
+        CHECK_BROKEN_LINKS.out.valid_genomes_ch
+            .filter { it != null }
+            .map{
+                [
+                    accno: it[0],
+                    genome_fna: it[1],
+                    genome_gff: it[2]
+                ]
+            }
+
     emit:
         gindex           = GENOMES_SKETCH.out.signatures
         sindex           = SAMPLES_SKETCH.out.signatures
-        filtered_genomes = ch_filtered_genomes
+        filtered_genomes = ch_valid_genomes_filtered
         versions         = ch_versions
 }