Transipedia / dekupl-run

Identify differentially expressed k-mers between RNA-Seq datasets
MIT License
11 stars 11 forks source link

Sort does not use specified temp directory #52

Closed amirkdv closed 5 years ago

amirkdv commented 5 years ago

Multiple rules in Snakefile use sort in their recipes:

rule jellyfish_dump:
  # truncated
  run:
    start_log(log['exec_time'], "jellyfish_dump")
    shell("{JELLYFISH_DUMP} -c {input} | {SORT} -k 1 -S {resources.ram}M --parallel {threads}| pigz -p {threads} -c > {output}")
    # truncated

and

rule ref_transcript_dump:
  # truncated
  run:
    start_log(log['exec_time'], "ref_transcript_dump")
    shell("{JELLYFISH_DUMP} -c {input} | {SORT} -k 1 -S {resources.ram}M --parallel {threads}| pigz -p {threads} -c > {output}")
    # truncated

The default temp directory for sort is /tmp which can be overridden via -T which means that with large input files the root filesystem can easily fill up and crash de-kupl.

Pull request to add the -T option to usages of sort will be sent shortly.

amirkdv commented 5 years ago

Note that the -T option is also respected by sort in OS X.