如何在Snakemake中为配置参数设置多组取值？

核心思路 #

将多组参数打包成字符串传入Snakefile，解析为参数组合列表后，通过expand遍历所有组合生成任务，规则通过wildcards捕获文件名中的参数值，自动适配每组参数运行。

修改后的Snakefile代码 #

import os
grouphome = os.environ['GROUPHOME']
ISOLATES = [i for i in open(grouphome+'/isolates.txt').read().split('\n') if len(i) > 0]

# 解析命令行传入的多组参数
param_sets = config.get("param_sets", "")
# 拆分参数组并转换为整数元组，过滤空字符串
PARAM_COMBOS = [tuple(map(int, set_str.split(','))) for set_str in param_sets.split(';') if set_str.strip()]
# 校验每组参数数量是否正确
assert all(len(combo) == 3 for combo in PARAM_COMBOS), "每组参数必须包含window_size、slide_size、mean_filter_size三个值"

rule all:
    input:
        # 生成所有参数组合+样本的预处理文件
        expand(
            grouphome+"/preprocessed_coverage_{ws}_{ss}_{mfs}/{isolate}.csv", 
            isolate=ISOLATES,
            ws=[c[0] for c in PARAM_COMBOS],
            ss=[c[1] for c in PARAM_COMBOS],
            mfs=[c[2] for c in PARAM_COMBOS],
            allow_missing=False
        ),
        # 生成所有参数组合+样本的ptr文件
        expand(
            grouphome+"/preprocessed_coverage_{ws}_{ss}_{mfs}/ptr/{isolate}/ptr.csv", 
            isolate=ISOLATES,
            ws=[c[0] for c in PARAM_COMBOS],
            ss=[c[1] for c in PARAM_COMBOS],
            mfs=[c[2] for c in PARAM_COMBOS],
            allow_missing=False
        ),
        # 生成所有参数组合+样本的残差拟合值文件
        expand(
            grouphome+"/preprocessed_coverage_{ws}_{ss}_{mfs}/ptr/{isolate}/residuals_fitted_values.csv",
            isolate=ISOLATES,
            ws=[c[0] for c in PARAM_COMBOS],
            ss=[c[1] for c in PARAM_COMBOS],
            mfs=[c[2] for c in PARAM_COMBOS],
            allow_missing=False
        )

rule coverage_preprocessing:
    input:
        input1= grouphome+"/data/{isolate}.fasta",
        input2= grouphome+"/max_coverage/{isolate}_maximum_coverage.csv"
    output: 
        grouphome+"/preprocessed_coverage_{ws}_{ss}_{mfs}/{isolate}.csv"
    shell:
        """./preprocess_coverage.py -i {input.input2} -g {input.input1} -sw {wildcards.ss} -ws {wildcards.ws} -m mean -bm mean -bs {wildcards.mfs} -o {output}"""

rule calc_bptr:
    input:
        input1= grouphome+"/data/{isolate}.fasta",
        input2= grouphome+"/preprocessed_coverage_{ws}_{ss}_{mfs}/{isolate}.csv"
    output:
        output1= grouphome+"/preprocessed_coverage_{ws}_{ss}_{mfs}/ptr/{isolate}/ptr.csv",
        output2= grouphome+"/preprocessed_coverage_{ws}_{ss}_{mfs}/ptr/{isolate}/residuals_fitted_values.csv",
    shell:
        """./calculate_bptr.py -i {input.input2} -b mean -s {wildcards.mfs} -g {input.input1} -f {output.output2} -o {output.output1}"""

命令行运行方式 #

直接通过param_sets参数传递多组参数，每组用分号分隔，组内三个参数用逗号分隔：

snakemake --config param_sets="1000000,10000,99;1000000,100,66" --cores=9

关键修改点说明 #

1. 参数解析：将命令行传入的字符串拆分为多个参数组合元组，确保每组包含三个参数
2. rule all扩展：使用expand遍历所有参数组合和样本，生成所有需要的输出文件路径
3. 规则适配：移除原规则中直接读取config单值的逻辑，改用wildcards从输出文件名中捕获对应参数值，自动适配每组参数运行

内容的提问来源于stack exchange，提问作者Mary