kkonganti@0: process MASHTREE {
kkonganti@0:     tag "$meta.id"
kkonganti@0:     label 'process_medium'
kkonganti@0: 
kkonganti@0:     module (params.enable_module ? "${params.swmodulepath}${params.fs}mashtree${params.fs}1.4.3" : null)
kkonganti@0:     conda (params.enable_conda ? "bioconda::mashtree=1.4.3 conda-forge::perl conda-forge::coreutils" : null)
kkonganti@0:     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
kkonganti@0:         'https://depot.galaxyproject.org/singularity/mashtree:1.4.3--pl5321h031d066_0' :
kkonganti@0:         'quay.io/biocontainers/mashtree:1.4.3--pl5321h031d066_0' }"
kkonganti@0: 
kkonganti@0:     input:
kkonganti@0:     tuple val(meta), path(seqs)
kkonganti@0:     path reference
kkonganti@0: 
kkonganti@0:     output:
kkonganti@0:     tuple val(meta), path("*.dnd"), emit: tree
kkonganti@0:     tuple val(meta), path("*.tsv"), emit: matrix
kkonganti@0:     path "versions.yml"           , emit: versions
kkonganti@0: 
kkonganti@0:     when:
kkonganti@0:     task.ext.when == null || task.ext.when
kkonganti@0: 
kkonganti@0:     script:
kkonganti@0:     def args      = task.ext.args ?: ''
kkonganti@0:     def prefix    = task.ext.prefix ?: "${meta.id}"
kkonganti@0:     def fofn      = (args.toString().matches(/.*file\-of\-files/) ? 'true' : 'false')
kkonganti@0:     def fas       = reference.collect().join('\\n')
kkonganti@0:     def ref_fasta = (fas ?: '') 
kkonganti@0:     """
kkonganti@0: 
kkonganti@0:     echo -e "$ref_fasta" | while read -r fa; do
kkonganti@0:         if ! grep -F "\${fa}" $seqs && [ "$fofn" = "true" ]; then 
kkonganti@0:             echo "\${fa}" >> $seqs
kkonganti@0:         fi
kkonganti@0:     done
kkonganti@0: 
kkonganti@0:     mashtree \\
kkonganti@0:         $args \\
kkonganti@0:         --tempdir "." \\
kkonganti@0:         --numcpus $task.cpus \\
kkonganti@0:         --outmatrix ${prefix}.tsv \\
kkonganti@0:         --outtree ${prefix}.dnd \\
kkonganti@0:         $seqs
kkonganti@0: 
kkonganti@0:     sed -ie 's/_scaffolded_genomic//g' ${prefix}.dnd
kkonganti@0: 
kkonganti@0:     cat <<-END_VERSIONS > versions.yml
kkonganti@0:     "${task.process}":
kkonganti@0:         mashtree: \$( echo \$( mashtree --version 2>&1 ) | sed 's/^.*Mashtree //' )
kkonganti@0:     END_VERSIONS
kkonganti@0: 
kkonganti@0:     sedver=""
kkonganti@0:     grepver=""
kkonganti@0: 
kkonganti@0:     if [ "${workflow.containerEngine}" != "null" ]; then
kkonganti@0:         sedver=\$( sed --help 2>&1 | sed -e '1!d; s/ (.*\$//' )
kkonganti@0:         grepver="\$sedver"
kkonganti@0:     else
kkonganti@0:         sedver=\$( echo \$(sed --version 2>&1) | sed 's/^.*(GNU sed) //; s/ Copyright.*\$//' )
kkonganti@0:         grepver=\$( echo \$(grep --version 2>&1) | sed 's/^.*(GNU grep) //; s/ Copyright.*\$//' )
kkonganti@0:     fi
kkonganti@0: 
kkonganti@0:     cat <<-END_VERSIONS >> versions.yml
kkonganti@0:         grep: \$grepver
kkonganti@0:         sed: \$sedver
kkonganti@0:     END_VERSIONS
kkonganti@0:     """
kkonganti@0: }