kkonganti@17: process BCS_DISTANCE_MATRIX {
kkonganti@17:     tag "Samples vs Genomes"
kkonganti@17:     label "process_pico"
kkonganti@17: 
kkonganti@17:     module (params.enable_module ? "${params.swmodulepath}${params.fs}python${params.fs}3.8.1" : null)
kkonganti@17:     conda (params.enable_conda ? "conda-forge::python=3.10 conda-forge::pyyaml" : null)
kkonganti@17:     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
kkonganti@17:         'https://depot.galaxyproject.org/singularity/multiqc:1.14--pyhdfd78af_0' :
kkonganti@17:         'quay.io/biocontainers/multiqc:1.14--pyhdfd78af_0' }"
kkonganti@17: 
kkonganti@17:     input:
kkonganti@17:         path matrix
kkonganti@17:         path labels
kkonganti@17: 
kkonganti@17:     output:
kkonganti@17:         path 'bcs_sourmash_matrix.tblsum.txt', emit: mqc_txt, optional: true
kkonganti@17:         path 'bcs_sourmash_matrix_mqc.json'  , emit: mqc_json, optional: true
kkonganti@17:         path 'bcs_sourmash_matrix_mqc.yml'   , emit: mqc_yml, optional: true
kkonganti@17:         path 'versions.yml'                  , emit: versions
kkonganti@17: 
kkonganti@17:     when:
kkonganti@17:         task.ext.when == null || task.ext.when
kkonganti@17: 
kkonganti@17:     script:
kkonganti@17:         def args = task.ext.args ?: ''
kkonganti@17: 
kkonganti@17:         """
kkonganti@17:         sourmash_sim_matrix.py \\
kkonganti@17:             -pickle ${params.tuspy_ps} \\
kkonganti@17:             -csv $matrix \\
kkonganti@17:             -labels $labels
kkonganti@17: 
kkonganti@17:         if [ -e "bcs_sourmash_matrix.tblsum.txt" ] && [ -s "bcs_sourmash_matrix.tblsum.txt" ]; then
kkonganti@17:             create_mqc_data_table.py \\
kkonganti@17:                 "bcs_sourmash_matrix" \\
kkonganti@17:                 "The following table is an asymmetrical matrix of all <code>v.</code> all <b>ANI</b> values between each of the sample <b>FASTQ</b>'s sketch and genome <b>FASTA</b>'s sketch. Please note that there may be multiple genomes belonging to a serotype as processed by the initial screening steps of the workflow."
kkonganti@17:         fi
kkonganti@17: 
kkonganti@17:         cat <<-END_VERSIONS > versions.yml
kkonganti@17:         "${task.process}":
kkonganti@17:             python: \$( python --version | sed 's/Python //g' )
kkonganti@17:             bash: \$( bash --version 2>&1 | sed '1!d; s/^.*version //; s/ (.*\$//' )
kkonganti@17:         END_VERSIONS
kkonganti@17:         """
kkonganti@17: }