BinChicken single run with 3 metagenomes did not finish after 72h: expected runtime or potential misconfiguration?

[xtan824]

Hi AroneyS,

I tried to run BinChicken single with --run-aviary on a HPC system, but the job did not complete within 72 hours and was finally killed due to walltime limit, even though i only have 3 paired-end metagenomic samples and allocated substantial resources (32 cores, 320 GB). I would like to ask whether this runtime is expected for BinChicken single (with --run-aviary) even with only three samples, or there might be an installation issue, suboptimal configuration, or incorrect usage?

Here are my job submission script and relevant log files for your reference.

binchicken.sh
binchicken_job.log
logs.zip

Look forward to hearing from you.

Best regards,
Xin

[AroneyS]

Hi Xin,

Yes, that is expected when using --run-aviary. Aviary will run assembly (often takes 6-24 hours per assembly, depending on sample complexity) and binning (often takes 2 days to 1 week per assembly, depending on how many samples you are using).

From the logs, it spent most of the time on assembly of one of the samples, then started binning for that same sample. Rerunning the same command will continue where it left off, so you could just keep retrying until it finishes. None of the individual jobs should take longer than 72h.

Otherwise, you may want to look into using a Snakemake profile to get Bin Chicken to submit individual jobs (e.g. assembly, separate jobs for binners, etc.) to your HPC. See --snakemake-profile option. This means each job will have separate custom CPU/RAM/walltime and should complete without you having to rerun the command.

Hope that helps,
Sam

[xtan824]

Hi Sam,

Thanks for your reply. I am trying to rerun the same command as you suggested.

Regards,
Xin