RnaSnake/Includes/Annotation.snake

######################################################
# Annotation with eggnog-mapper
######################################################
Annotation_Input = config["Annotation_Input"]
Annotation_Dir = config["Annotation_Dir"]
EMAPPER_HOME = config["EMAPPER_HOME"]
diamond_nr_db = config["diamond_nr_db"]
blast_method = "blastp"
emapper_method = " "

if config["data_type"] == "nucl":
    blast_method = "blastx"
    emapper_method = " --translate "


rule emapper_search:
    input:
        Annotation_Input
    output:
        Annotation_Dir + "/my.emapper.seed_orthologs"
    threads: 32
    shell:
        "export PATH={EMAPPER_HOME}/bin:$PATH;"
        "python2 {EMAPPER_HOME}/emapper.py -m diamond  -i {input} -o {Annotation_Dir}/my --cpu {threads} --no_annot --no_file_comments {emapper_method}"

rule emapper_annot:
    input:
        Annotation_Dir + "/my.emapper.seed_orthologs"
    output:
        Annotation_Dir + "/my.emapper.annotations"
    threads: 32
    shell:
        "export PATH={EMAPPER_HOME}/bin:$PATH;"
        "python2 {EMAPPER_HOME}/emapper.py --annotate_hits_table {input} -o {Annotation_Dir}/my --cpu {threads} "

rule makeOrgDB:
    input:
        emapper = Annotation_Dir + "/my.emapper.annotations",
        Annotation_Input = Annotation_Input
    output:
        Annotation_Dir + "/anno_stat.txt"
    shell:
        "cd {Annotation_Dir} ;"
        "Rscript ../tools/emcp/makeOrgPackageFromEmapper.R my.emapper.annotations;"
        "Rscript ../tools/emcp/AnnoStat.R ../{input.Annotation_Input};"
        "cd ..;"

rule nr:
    input:
        Annotation_Input
    output:
        Annotation_Dir + "/my.nr.diamond.tab"
    threads:
        20
    shell:
        "diamond {blast_method} --threads {threads} --evalue 1e-10 --outfmt 6 -o {output} --query {input} --db {diamond_nr_db} --max-target-seqs 5"