comparative-genomics-project/workflow/modules/clustering.nf

process BLASTP_TO_ABC {

    input:
    path blastp

    output:
    path 'graph.abc'

    script:
    """
    awk 'BEGIN { OFS="\t" } { print \$14, \$16, \$12 }' "${blastp}" > 'graph.abc'
    """
}

process MCL {

    input:
    path abc

    output:
    path 'clustering.mcl'

    script:
    """
    mcl "${abc}" --abc -o 'custering.mcl'
    """
}

process MCL_TO_TSV {

    input:
    path mcl

    output:
    path 'families.tsv'

    script:
    """
    awk -f "${baseDir}/scripts/mcl_to_tsv.awk" "${mcl}" > 'families.tsv'
    """
}

workflow CLUSTERING {
    take:
    blastp_tsv

    main:
    BLASTP_TO_ABC(blastp_tsv)
    MCL(BLASTP_TO_ABC).out
    MCL_TO_TSV(MCL.out)

    emit:
    families = MCL_TO_TSV.out
}