Snakefile

rule all:
    input:
        "data/diamond.dmnd",


rule fetch_references:
    input:
        config=config["config_file"],
    output:
        reference_fasta="data/reference_genomes.fasta",
        reference_proteins="data/reference_proteins.fasta",
    shell:
        """
        python3 scripts/fetch_reference_genomes.py \
            --config-file {input.config} --reference-nucleotides-file {output.reference_fasta} \
            --reference-proteins-file {output.reference_proteins}
        """


rule make_diamond_db:
    input:
        reference_fasta="data/reference_proteins.fasta",
    output:
        diamond_db="data/diamond.dmnd",
    shell:
        """
        diamond makedb \
            --in {input.reference_fasta} \
            -d {output.diamond_db}
        """