Humann3.8 --resume doesn't work

Microbial community profiling HUMAnN
1

Hi,

I ran humann with --resume option in hope that it can use files already created by previous aborted run. However, it created a second temp directory and ignored the files in the first temp directory. Here is my command line:
`

humann --input $work_dir/$sampleID/kneaddata/$sampleID.fastq --input-format fastq --taxonomic-profile $work_dir/$sampleID/metaphlan/${sampleID}.taxonomy_profile.txt --output $work_dir/$sampleID/humann --threads $thread_num --remove-temp-output --o-log $work_dir/$sampleID/humann/$sampleID.humann.log --output-basename $sampleID --diamond-options “–block-size 3.0” --resume

`
Here is the output files:

humann_result
humann_result738×734 51.2 KB

Is there something I did in a wrong way?

2

After looking at the code, I think the reason is that I set ‘–remove-temp-output’ option in which case the program will create a temp directory with random character string as suffix instead of using a fixed directory name. Here is related codes:

# set the location of the temp directory
    if not args.remove_temp_output:
        config.temp_dir=os.path.join(output_dir,config.file_basename+"_humann_temp")
        if not os.path.isdir(config.temp_dir):
            try:
                os.mkdir(config.temp_dir)
            except EnvironmentError:
                sys.exit("Unable to create temp directory: " + config.temp_dir)
    else:
        config.temp_dir=tempfile.mkdtemp(
            prefix=config.file_basename+'_humann_temp_',dir=output_dir)
3

Yes, you are totally right. If you don’t have the intermediate outputs, HUMAnN is not able to run in the “–resume” mode as it does not have the data files it needs to skip steps.

Thanks for answering your own question and quickly too!
Lauren