有没有办法为ruleorder提供函数？

Question

是否可以为ruleorder提供一个函数，像这样？

def determine_ruleorder(wildcards, config):
    samples_file = pd.read_table(config['sample_file'], sep='\t')
    sample_layout = samples_file.loc[samples_file['sample'] == wildcards.sra, 'layout'].values[0]
    
    if sample_layout == "single":
        return "fasterq_dump_se > fasterq_dump_pe"
    elif sample_layout == "paired":
        return "fasterq_dump_pe > fasterq_dump_se"
    else:
        raise ValueError(f"Unknown sample type for SRA '{wildcards.sra}': {sample_layout}")

ruleorder: determine_ruleorder

在这种情况下，我收到错误：

UnknownRuleException:
Error in ruleorder definition. There is no rule named determine_ruleorder

这将是示例文件：

sample  layout
SRR6308266  paired
SRR1525658  single
SRR6308195  paired
SRR6206507  paired
SRR22957809 paired
SRR22957815 paired
SRR21169022 paired
SRR13066396 paired
SRR7904254  paired
SRR6308265  paired
SRR1525659  single
SRR1525660  single

Answer 1

好像没有。规则顺序会影响工作流程的全局执行，而输入函数需要单个通配符来评估，因此它似乎不太可能被实现。

但是，在这种情况下您不需要这样做。如果输入函数引发值错误，则尝试下一个合理的规则：

ruleorder:
    dump_se > dump_pe

def dump_se_input(wildcards):
    samples_file = pd.read_table(config['sample_file'], sep='\t')
    sample_layout = samples_file.loc[samples_file['sample'] == wildcards.sra, 'layout'].values[0]
    
    if sample_layout == "single":
        return FILES
    raise ValueError("May be paired")

rule dump_se:
    input: dump_se_input
    ...

def dump_pe_input(wildcards):
    samples_file = pd.read_table(config['sample_file'], sep='\t')
    sample_layout = samples_file.loc[samples_file['sample'] == wildcards.sra, 'layout'].values[0]
    
    if sample_layout == "paired":
        return FILES
    raise ValueError(f"Unknown sample type for SRA '{wildcards.sra}': {sample_layout}")

rule dump_pe:
    input: dump_pe_input
    ...

因此首先根据 SE 规则评估样本。如果它们配对，则值错误会导致 PE 规则运行。它还进行了一项检查，该检查将引发（未捕获的）错误，警告用户样本不符合要求。