annotate Snakemake_files/Snakefile_cactus_heatmap_upset @ 10:d103c41b6931 draft

Uploaded

import glob
import os
import shutil

import yaml
configfile: "config.yaml"


SAMPLES = []
with open("config.yaml", "r") as yaml_file:
    genome_data = yaml.safe_load(yaml_file)
    for genome_name, genome_info in genome_data["input_genomes"].items():
        name = genome_info["name"]
        SAMPLES.append(name)


rule all:
    input:
        "outputs/seqfile",
        "outputs/cactus_outdir/output_cactus.sv.gfa.gz",
        "outputs/pav_matrix.tsv",
        "outputs/heatmap.svg.gz",
        "outputs/rarefaction_curves.txt"

rule get_data:
    input:
        "config.yaml"
    output:
        "outputs/seqfile",
        expand("outputs/genomes/{sample}.fasta", sample=SAMPLES),
        expand("outputs/genomes/{sample}.gff", sample=SAMPLES),
    run:
        with open(input[0], "r") as yaml_file, open(output[0], "w") as output_file:
            genome_data = yaml.safe_load(yaml_file)
            for genome_name, genome_info in genome_data["input_genomes"].items():
                fasta_file = genome_info["fasta"]
                name = genome_info["name"]
                gff_file = genome_info["gff3"]
                shutil.copy(gff_file, "outputs/genomes/"+str(name)+".gff")
                shutil.copy(fasta_file, "outputs/genomes/"+str(name)+".fasta")
                output_file.write(f"{genome_name}\toutputs/genomes/"+str(name)+".fasta\n")


rule gffread:
    input:
        fasta = "outputs/genomes/{sample}.fasta",
        gff = "outputs/genomes/{sample}.gff"
    output:
        "outputs/genomes/{sample}.nuc"
    shell:
        """
        gffread -x {output} -g {input.fasta} {input.gff}
        """

rule cactus:
    input:
        expand("outputs/genomes/{sample}.fasta", sample=SAMPLES),
        expand("outputs/genomes/{sample}.gff", sample=SAMPLES),
    params:
        reference=SAMPLES[0]
    output:
        gfasv="outputs/cactus_outdir/output_cactus.sv.gfa.gz",
        gfafull="outputs/cactus_outdir/full.gfa"
    shell:
        """
        mkdir outputs/cactus_work
        cactus-pangenome outputs/jobstore outputs/seqfile --outDir outputs/cactus_outdir --outName output_cactus --reference {params.reference} --workDir outputs/cactus_work --vcf --mapCores 12 --gfa clip full --viz
        zcat outputs/cactus_outdir/output_cactus.full.gfa.gz >{output.gfafull}
        """


rule gfatools:
    input:
        "outputs/cactus_outdir/output_cactus.sv.gfa.gz"
    output:
        "outputs/cactus_outdir/output_cactus.sv.gfa.pangenome.fasta"
    shell:
        """
        gfatools gfa2fa -s {input} > {output}
        """

rule minigraph:
    input:
        cds_fasta="outputs/genomes/{sample}.nuc",
        gfa="outputs/cactus_outdir/output_cactus.sv.gfa.gz",
    output:
        gaf="outputs/genomes/{sample}.nuc.pangenome.gaf",
    shell:
        """
        minigraph -x lr {input.gfa} {input.cds_fasta} > {output.gaf}
        """

rule odgi:
    input:
        "outputs/cactus_outdir/full.gfa"
    output:
        og="outputs/cactus_outdir/full.gfa.og",
        png="outputs/cactus_outdir/full.gfa.png",
    shell:
        """
        odgi build -g {input} -o {output.og}
        odgi viz -i {output.og} -o {output.png}
        """

rule gaf2bed:
    input:
        "outputs/genomes/{sample}.nuc.pangenome.gaf"
    output:
        "outputs/genomes/{sample}.nuc.pangenome.gaf.bed",
    shell:
        """
        awk '{{print $6"\\t"$8"\\t"$9"\\t"$1}}' {input} >{output}
        """

rule bedtools_intersect:
    input:
        strains="outputs/seqfile",
        bedfiles=expand("outputs/genomes/{sample}.nuc.pangenome.gaf.bed", sample=SAMPLES)
    output:
        "outputs/pav_matrix.tsv",
    shell:
        """
        perl $PANEX_PATH/Perl/GeneratePAVfromBed.pl {input.strains} outputs/genomes {output}
        """

rule heatmap_upset:
    input:
        pav="outputs/pav_matrix.tsv"
    output:
        heatmap="outputs/heatmap.svg.gz",
        html="outputs/heatmap.svg.heatmap_plotly.html",
        upsetr="outputs/upsetr.svg",
        binpav="outputs/heatmap.svg.pangenes_01matrix.txt"
    shell:
        """
        perl $PANEX_PATH/Perl/GenerateHeatmapFromPAV.pl {input.pav} outputs/heatmap.svg
        mv outputs/heatmap.svg.upsetr.svg {output.upsetr}
        """

rule micropan:
    input:
        binpav="outputs/heatmap.svg.pangenes_01matrix.txt"
    output:
        txt="outputs/rarefaction_curves.txt",
        pdf="outputs/rarefaction_curves.pdf",
        svg="outputs/rarefaction_curves.svg",
        heaps="outputs/heaps.tsv"
    shell:
        """
        Rscript $PANEX_PATH/R/micropan_rarefaction.R -f {input.binpav} -p {output.pdf} -a {output.heaps} -o {output.txt}
        pdf2svg {output.pdf} {output.svg}
        """