Snakefile1

# Snakefile

from read_accessions import read_txt_file

accession_list = read_txt_file("accessions.txt")

print("Accession list:", accession_list)

rule all:
    input:
        expand("snakemake/{sample}/{sample}_fastqc.html", sample=accession_list),
        "multiqc_report.html"

rule download_sra:
    output:
        "snakemake/{sample}/{sample}.sra"
    params:
        sra_id=lambda wildcards: wildcards.sample
    shell:
        """
        echo 'Running download_sra for {wildcards.sample}' &&
        prefetch --output-directory snakemake/{params.sra_id} --output-file snakemake/{params.sra_id}/{params.sra_id}.sra {params.sra_id} &&
        [ ! -e snakemake/{params.sra_id}/{params.sra_id}.sra ] || [ "snakemake/{params.sra_id}/{params.sra_id}.sra" -ef "{output}" ] || mv snakemake/{params.sra_id}/{params.sra_id}.sra {output} ||
        (echo 'Error in download_sra' && exit 1)
        """

rule fastq_dump:
    input:
        "snakemake/{sample}/{sample}.sra"
    output:
        "snakemake/{sample}/{sample}.fastq"
    shell:
        """
        echo 'Running fastq_dump for {wildcards.sample}' &&
        fastq-dump --outdir snakemake/{wildcards.sample} {input} &&
        [ ! -e snakemake/{wildcards.sample}/{wildcards.sample}.fastq ] || [ "snakemake/{wildcards.sample}/{wildcards.sample}.fastq" -ef "{output}" ] || mv snakemake/{wildcards.sample}/{wildcards.sample}.fastq {output} ||
        (echo 'Error in fastq_dump' && exit 1)
        """

rule fastqc_raw:
    input:
        "snakemake/{sample}/{sample}.fastq"
    output:
        html="snakemake/{sample}/{sample}_fastqc.html",
        zip="snakemake/{sample}/{sample}_fastqc.zip"
    shell:
        """
        echo "Processing sample: {wildcards.sample}"
        fastqc {input}
        """

rule multiqc:
    input:
        expand("snakemake/{sample}/{sample}_fastqc.html", sample=reversed(accession_list))
    output:
        "multiqc_report.html"
    shell:
        "multiqc . --filename {output} -f"