/**
/** Comparative Genomics workflow
/**
/** This workflow find the duplicate genes from a proteome
/** Then, It finds the Tandemly Arrayed Genes (TAGs)
/**/

nextflow.enable.dsl = 2;

include { GUNZIP } from "./modules/gunzip.nf"
include { BLAST_ALL_AGAINST_ALL } from "./modules/blast.nf"
include { FILTER_FASTA } from "./modules/filter_fasta.nf"
include { FILTER_BLASTP } from "./modules/filter_blastp.nf"
include { CLUSTERING } from "./modules/clustering.nf"

process PROTEIN_GENE_MAPPING {

    input:
    path proteome

    output:
    path 'protein_gene.tsv'
}

workflow {
    proteome = Channel.fromPath(params.proteome)
    GUNZIP(proteome)
    FILTER_FASTA(GUNZIP.out)
    BLAST_ALL_AGAINST_ALL(FILTER_FASTA.out.proteome)
    FILTER_BLASTP(params.min_coverage, params.min_identity, BLAST_ALL_AGAINST_ALL.out, FILTER_FASTA.out.lengths)

    CLUSTERING(FILTER_BLASTP.out)
}