How to use Snakemake with tools which don't have clearly defined output options for files?

Question

[Artem Kiselev]

Hi, I would like to use Snakemake in my variant annotation pipeline. But Annovar don't have clearly defined file output option (I can assign only output folder and annovar going to attach to filename: "multianno.hg19), I can't describe final annovar output in "rule all", since it differs from the output parameters in rule for annovar tool. Help, please solve this problem. Thank you!

Answer 1

[WouterDeCoster]

You specify output files that are created, and use a params value for your shell code.

Example below for mosdepth:
I specify that mosdepth/{sample}.mosdepth.global.dist.txt will be created, but as output value for my shell command I use mosdepth/{params.prefix}, which takes the sample name.
Mosdepth will create mosdepth/{sample}.mosdepth.global.dist.txt

rule mosdepth_get:
    input:
        bam = "alignment/{sample}.bam",
        bai = "alignment/{sample}.bam.bai"
    output:
        protected("mosdepth/{sample}.mosdepth.global.dist.txt"),
        protected("mosdepth/{sample}.regions.bed.gz"),
    params:
        windowsize = 500,
        prefix = "{sample}",
    log:
        "logs/{aligner}/mosdepth/mosdepth_{sample}.log"
    shell:
        "mosdepth -n --by {params.windowsize} mosdepth/{params.prefix} {input.bam} 2> {log}"