Megahit outputs blank final.contigs.fa

[aimirza]

Megahit created a blank final.contigs.fa file on my smallest sample (2 million reads). final.contigs.fa.300fa file is missing. I don't have issues with other samples. The job ran with no apparent problems it seems. The log says "Merging to output final contigs." All the intermediate_contigs are there. Here is the assembly log file. Any ideas?

log.txt

[aimirza]

For some reason, it doesn't want to merge the intermediate contigs. I reran assembly with --rerun-incomplete parameter after deleting the empty final.contigs.fa file from the sunbeam_output/assembly/megahit/{sample}_asm directory hoping it would merge them. Didn't work, it just ran paired megahit again and still outputted a blank final.contigs.fa. Finally, after deleting the whole sunbeam_output/assembly/megahit/{sample}_asm directory it worked. Not sure why this issue happens. Maybe change sunbeam to not rerun megahit if the intermediate files are present and have a merging step.

[louiejtaylor]

Thanks for the issue, and sorry for the slow response!

Hmm. It sounds like something failed in MEGAHIT during the initial running of the rule, in a way that impeded subsequent attempts to build contigs for that sample. As far as I can tell from the log file, it looks like MEGAHIT thought it ran fine--which is extra confusing.

As far as fixing the problem, you did exactly what I would've suggested--deleting the intermediate directory and rerunning the sample. Maybe we can explicitly state this in the rule, although I'm not sure it'll help, since it sounds like the rule ran fine (but just created an empty contigs file). Do you mind sharing the version of MEGAHIT you're running so I can try to reproduce this and hopefully avoid people having this problem in the future?

[aimirza]

I used MEGAHIT v1.1.3

I couldn't find a similar issue in the megahit github repository. I wonder if to rerun Megahit the intermediate files must be deleted first for megahit to properly run.

[evolphotonics]

@louiejtaylor, I am facing the same issue with Megahit v1.2.9. any help much appreciated attaching the logs... megahit_run_log.txt megahit.nohup.output.txt options.json.txt

[Ulthran]

@evolphotonics I notice force_overwrite is false, I wonder if setting that to true would achieve the desired effect of automatically overwriting intermediate files. Unfortunately I'm not sure there's anything to be done about it failing silently sometimes, but force_overwrite might make it easier to rerun in those cases.

[evolphotonics]

thanks - I got it to work in the end...it did not happen when I ran it again or when I changed some parameters, but good to know about force_overwrite

On Thu, Sep 29, 2022 at 7:51 PM Charlie @.***> wrote:

@evolphotonics https://github.com/evolphotonics I notice force_overwrite is false, I wonder if setting that to true would achieve the desired effect of automatically overwriting intermediate files. Unfortunately I'm not sure there's anything to be done about it failing silently sometimes, but force_overwrite might make it easier to rerun in those cases.

— Reply to this email directly, view it on GitHub https://github.com/sunbeam-labs/sunbeam/issues/271#issuecomment-1262357058, or unsubscribe https://github.com/notifications/unsubscribe-auth/AU2ERMMB4S77QQPQGQYOY3LWAWQX3ANCNFSM4KVTVKKQ . You are receiving this because you were mentioned.Message ID: @.***>

[Ulthran]

Great, thanks for the reply. I'm going to close this issue, please open a new one if this kind of behavior is observed in sunbeam3.