Snakemake调用Docker容器执行vcf2maf时文件找不到问题求助

Let's break down the root causes of your file-not-found error and fix them step by step:

1. Mismatched Local vs. Container Filesystem Paths #

The core issue is that you're passing local filesystem paths (from snakemake.input['read'] and snakemake.output['out']) directly to the Docker container. The container can't access your local files unless they're covered by the -v volume mounts you defined.

Your local directory /home/data/TEST/ is mapped to /data3 inside the container, so:

• Local input path: /home/data/TEST/CALL/{sample}.muetect2.vt.clean.vcf.gz
• Container equivalent input path: /data3/CALL/{sample}.muetect2.vt.clean.vcf.gz
• Local output path: /home/data/TEST/MAF/{sample}.muetect2.vt.vep.maf
• Container equivalent output path: /data3/MAF/{sample}.muetect2.vt.vep.maf

You need to convert these local paths to their container-side equivalents before passing them to the vcf2maf command.

2. Incorrect vcf2maf.pl Parameter Syntax #

You’re passing the output file directly after the input VCF, but vcf2maf.pl requires the explicit --output-maf flag to specify the output location. Omitting this flag makes the script misinterpret your arguments, leading to unexpected file lookup failures.

3. Docker Interactive Mode Conflict #

Using -it in your Docker command is designed for interactive terminal sessions. When running from a non-interactive Snakemake script, this can cause hangs or silent failures. Replace -it with --rm to automatically clean up the container after it finishes running (a best practice for ephemeral jobs).

4. Wrong VEP Data Path #

Your --vep-data parameter is set to /mnt/, but you mapped your VEP database to /mnt/homo_sapiens inside the container. You need to point --vep-data to this correct container-side path.

Fixed report.py Script #

from subprocess import run

# Convert local paths to container-side paths using the volume mapping
container_input = snakemake.input['read'].replace("/home/data/TEST/", "/data3/")
container_output = snakemake.output['out'].replace("/home/data/TEST/", "/data3/")

cmd = [
    'docker', 'container', 'run', '--rm',  # Replace -it with --rm for non-interactive use
    '-v', snakemake.params['db_ens'],
    '-v', snakemake.params['data'],
    'vcf2maf_87',
    'perl', 'vcf2maf.pl',
    '--input-vcf', container_input,
    '--output-maf', container_output,  # Explicitly define output with --output-maf
    '--vep-data', '/mnt/homo_sapiens',  # Correct VEP data path in container
    '--ref-fasta', snakemake.params['fst'],
    '--tumor-id', snakemake.params['tumor_id'],
    '--normal-id', 'NORMAL'
]

# Add check=True to fail fast if the command encounters an error (easier debugging)
run(cmd, check=True)

Additional Debugging Tips #

• Ensure the local output directory /home/data/TEST/MAF/ exists before running Snakemake. You can add this to your rule to auto-create it:

  rule create_maf:
      # ... existing input/output/params ...
      shell:
          "mkdir -p /home/data/TEST/MAF/"
      script: "scripts/report.py"
  

• Verify that your VEP reference fasta path (/mnt/homo_sapiens/87_GRCh37/Homo_sapiens.GRCh37.75.dna.primary_assembly.fa) exists inside the container by running a quick test command:

  docker run --rm -v /mnt/mpwork/vep/homo_sapiens:/mnt/homo_sapiens vcf2maf_87 ls /mnt/homo_sapiens/87_GRCh37/
  

内容的提问来源于stack exchange，提问作者mau_who