使snakemake用作输入，2输出具有不同名称的规则

Question

使snakemake用作输入，2输出具有不同名称的规则

我正在制作一条蛇形管道，其中我使用strelka来比较肿瘤和正常样本。在这种情况下，我想比较GERMLINE = ("PT1", "S6", "S1”)的第一个元素和肿瘤TUMOR = ("T5", "T7", "T20")的第一个元素。

• PT1对T5
• S6对T7
• S1对T20

该管道适用于初始规则:文件夹、strelkaconfig和strelkarun。问题在于strelka输出的后处理，因为我想对这两个输出进行相同的处理：

• somatic.snvs.vcf
• somatic.indels.vcf

然而，我不知道如何让snakemake明白，在不重复规则的情况下，它应该对两者都做同样的事情。我试着做以下几件事：

GERMLINE = ("PT1", "S6", "S1")
TUMOR = ("T5", "T7", "T20")
ANALYSIS = "OUTPUT_PATH"
TYPEVAR = ["snvs","indels"]
INDGATK = "ref"

rule all:
    input:
        [ANALYSIS +"/{}_vs_{}/Stelka/results/variants/somatic.snvs.vcf".format(sample_g, sample_t)
         for (sample_g, sample_t) in zip(GERMLINE, TUMOR)],
        [ANALYSIS +"/{}_vs_{}/Stelka/runWorkflow.py".format(sample_g, sample_t)
         for (sample_g, sample_t) in zip(GERMLINE, TUMOR)],
        [ANALYSIS +"/{}_vs_{}/Stelka/results/variants/somatic.{}_Filtered".format(sample_g, sample_t,typevar)
         for (sample_g, sample_t,typevar)
         in zip(GERMLINE*len(TUMOR), TUMOR*len(TUMOR),sorted(TYPEVAR*len(TUMOR)))]


# Make folders
rule folders:
    input:
        g = "{samples_g}.bam",
        t = "{samples_t}.bam"
    output: 
        gen = "/{samples_g}_vs_{samples_t}",
        strelka = "/{samples_g}_vs_{samples_t}/Stelka/"
    run: 
        '''mkdir {output.gen}
        mkdir {output.strelka}'''

# Strelka configuration
rule strelkaconfig:
    input:
        g = "{samples_g}.bam",
        t = "{samples_t}.bam",
        out_dir = ANALYSIS + "/{samples_g}_vs_{samples_t}/Stelka/"
    output:
        wfs = ANALYSIS + "/{samples_g}_vs_{samples_t}/Stelka/runWorkflow.py"
    params:
        ref = INDGATK
    shell:
        "python configureStrelkaSomaticWorkflow.py --normalBam {input.g} --tumorBam {input.t} --referenceFasta {params.ref} --runDir {input.out_dir} "

# Strelka run
rule strelkarun:
    input:
        wfs = ANALYSIS + "/{samples_g}_vs_{samples_t}/Stelka/runWorkflow.py"
    output:
        outsnvs = ANALYSIS + "/{samples_g}_vs_{samples_t}/Stelka/results/variants/somatic.snvs.vcf",
        outindels = ANALYSIS + "/{samples_g}_vs_{samples_t}/Stelka/results/variants/somatic.indels.vcf"
    shell:
        "python {input.wfs}"

# POSTPROCESSING
rule vcfp:
    input: ANALYSIS + "/{samples_g}_vs_{samples_t}/Stelka/results/variants/somatic.{typevar}.vcf"
    output: ANALYSIS + "/{samples_g}_vs_{samples_t}/Stelka/results/variants/somatic.{typevar}_Filtered.vcf"
    shell:
        "java -jar StrelkaVCFParser -v {input} "

但是当我尝试运行时，我会得到这样的错误：

MissingInputException in line 15 of pipe:
Missing input files for rule folders:
T7/Stelka/results/variants/somatic.indels_Filtered.bam

Answer 1

这条管道似乎在推断通配符不是你所期望的那样。

您可以尝试使用通配符约束，如下所示：

wildcard_constraints:
    samples_g = "|".join(GERMLINE)
    samples_t = "|".join(TUMOR)

这可能解决不了您的问题，但是all规则的第三个输入在我看来并不是很清楚。您可以使用两个连续的expand (第一个使用zip )实现相同的目标，如下所示：

expand(expand(
    ANALYSIS + "/{sample_g}_vs_{sample_t}/Stelka/results/variants/somatic.{{typevar}}_Filtered",
    zip, sample_g=GERMLINE, sample_t=TUMOR), typevar=TYPEVAR)

注意typevar周围的双大括号，这样它就不会在第一个展开时展开。

在执行python3之后，您可以在from snakemake.io import expand解释器中测试这一点。

我个人认为如果更容易理解。