Minia bug: max rehashes reached: 65536000 (notify a developer)

Question

[fungopoo]

Hi,

Minia just crashed, error message reads,

[DSK: Pass 1/1, Step 2: counting kmers  ]  51.1 %   elapsed:   0 min 60 sec    estimated remaining:   0 min 57 sec   cpu:  760.5 %   mem: [ 127,  127,    0] MB EXCEPTION: OAHash: max rehashes reached: 65536000 (notify a developer)

Any suggestions as to how I may proceed?

This has happened on two independent runs of a 4.8 gb dataset.

Thanks,
Angus

[george.ry]

I suggest you do as it says, and notify one of the developers ...

They will probably also be able to best advise you on how to proceed.

[Devon Ryan]

FYI, the minia authors follow the "minia" tag here (a number of tools use biostars as their official support forum).

[Rayan Chikhi]

Indeed, we do. Biostar is a good way to reach us.

[Rayan Chikhi]

Is this the latest minia 2.0.3?

[fungopoo]

Hi,

Thanks, yes 2.0.3

I used kmergenie to find kmer size, and then -abundance-min 3, I have subsequently re-run it and it still crashes.

[Rayan Chikhi]

• What's special about this input dataset.. is it whole-genome DNA Illumina data?
• What's your system? (amount of memory, free disk space on working directory)
• Please try adding -max-memory 4000 to the command line, this might fix it for now
• The ideal way for me to debug this problem would be if you could send me the whole dataset (compressed with gzip), or a subsample where the bug still occurs, to [e-mail]

[fungopoo]

Hi,

A final thanks

Please try adding -max-memory 4000 to the command line, this might fix it for now

This worked like a charm. I have 16gb ram, quad-core 64 bit cpu, OSX 10.10 and it ran for 838 s; thanks for you advice and patience.

[Rayan Chikhi]

Good to know it worked.

If anyone else experiences it and sees this page, please let me know that it happened to you too, so that we'll know that it might not be an isolated case.

[marie.kraus7]

Hi, I experienced the same problem, Minia crashed a few times because of memory allocation. The above advice helped :-) I tried to assemble draft genome of not known size, one lane from HiSeq 125PE (2 files 80 GB unzipped) on workstation with 189GB memory, 19x2.4GHz running Ubuntu 14.04. Thanks for advice Marie Kraus

[Rayan Chikhi]

Thanks for letting us know :) We're preparing a next release of minia and hopefully it will fix that bug.

Answer 1

[fungopoo]

OK,

Trying -max-memory 4000; if it crashes again I'll get the data to you.

The data is low-coverage whole genome DNA, Illumina miseq platform. I used Galaxy to clean it up, join the paired ends and concatenate two separate runs.