FASTQ cleanup using Snakemake

Snakefile

PATH = '/home/rpetit/staphopia/apps/analysis-pipeline'
sample = 'N315_R1'

rule all:
    input:
        'analyses/fastq-stats/{sample}.cleanup.fastq.json'.format(
            sample=sample
        )

rule bbduk_phix:
    input:
        fq = '{sample}.fastq.gz'.format(sample=sample)
    output:
        phix = 'analyses/fastq-qc/nophix_r1.fastq',
        json = 'analyses/fastq-stats/{sample}.original.fastq.json'.format(
            sample=sample
        )
    benchmark:
        "logs/benchmarks/bbduk-phix.txt"
    shell:
        "zcat {input.fq} | {PATH}/bin/fastq-stats > {output.json};"
        "{PATH}/bin/bbduk.sh -Xmx2g k=31 threads=1 in={input.fq} "
        "out={output.phix} stats=logs/fastq_cleanup-bbduk-phix.txt  hdist=1 "
        "ref={PATH}/data/fastq-qc/phiX-NC_001422.fasta  overwrite=t ordered=t;"

rule bbduk_adapter:
    input:
        fq = rules.bbduk_phix.output.phix
    output:
        noadapt = 'analyses/fastq-qc/noadapter_r1.fastq'
    benchmark:
        "logs/benchmarks/bbduk-adapter.txt"
    shell:
        "{PATH}/bin/bbduk.sh -Xmx2g threads=1 in={input.fq} "
        "out={output.noadapt} stats=logs/fastq_cleanup-bbduk-adapter.txt "
        "ref={PATH}/data/fastq-qc/adapters.fasta ktrim=r k=23 mink=11 hdist=1 "
        "tpe tbo qout=33 minlength=36 overwrite=t ordered=t"

rule spades_ec:
    input:
        fq = rules.bbduk_adapter.output.noadapt
    output:
        ec_fq = 'analyses/fastq-qc/corrected/noadapter_r1.00.0_0.cor.fastq',
        json = 'analyses/fastq-stats/{sample}.post-adapter.fastq.json'.format(
            sample=sample
        )
    benchmark:
        "logs/benchmarks/spades-ec.txt"
    shell:
        "cat {input.fq} | {PATH}/bin/fastq-stats > {output.json};"
        "{PATH}/bin/spades.py -s {input.fq} --only-error-correction "
        "--disable-gzip-output -t 1 -o analyses/fastq-qc/"

rule spades_stats:
    input:
        fq = rules.spades_ec.output.ec_fq
    output:
        json = 'analyses/fastq-stats/{sample}.post-ecc.fastq.json'
    shell:
        'cat {input.fq} | {PATH}/bin/fastq-stats > {output.json}'

rule fastq_cleanup:
    input:
        fq = rules.spades_ec.output.ec_fq,
        stats = 'analyses/fastq-stats/{sample}.post-ecc.fastq.json'
    output:
        final_fq = '{sample}.cleanup.fastq.gz'
    benchmark:
        "logs/benchmarks/fastq-clean.txt"
    shell:
        "cat {input.fq} | {PATH}/bin/fastq_cleanup --stats {input.stats} "
        "--coverage 100 | gzip --best - > {output.final_fq}"

rule final_stats:
    input:
        fq = rules.fastq_cleanup.output.final_fq
    output:
        json = 'analyses/fastq-stats/{sample}.cleanup.fastq.json'
    shell:
        "zcat {input.fq} | {PATH}/bin/fastq-stats > {output.json};"
        "rm -rf analyses/fastq-qc/"