oases memory

[RyNkA]

I was trying to run velvet/oases with kmer from 25 toward 67 on my local HPC, setting it to run with 256GB of ram. Velveth/g performed well, while oases stopped after around 10 hrs and displayed an error.

job killed: mem 269731380kb exceeded limit 268435456kb

My command was oases transcripts_69 -min_trans_lgth 200...Any suggestion ?

[shunyip]

Many HPCs have a setting that limits the maximum amount of ram that you can use in a job. If your job exceed that amount, it will be killed. In this case, it is 268435456kb.

It appears that this oases job requires more ram than your HPC allows. You should contact your HPC's administrator to see if you can gain access to more ram for this job; or you can find a way to decrease ram usage.

[RyNkA]

Many HPCs have a setting that limits the maximum amount of ram that you can use in a job. If your job exceed that amount, it will be killed. In this case, it is 268435456kb.

It appears that this oases job requires more ram than your HPC allows. You should contact your HPC's administrator to see if you can gain access to more ram for this job; or you can find a way to decrease ram usage.

Yes, I'm pretty well aware that Oases did actually exceed the amount of RAM allocated by the HPC. However i would like to ask if there is anyway to decrease ram usage. I've dig some search and they suggest that I lower the kmer value (which i did manage to do it at 31 kmer, however, the problem still presists)

Cheers,

[shunyip]

At the FAQ section of Oases' manual, they suggested several more ways for you to decrease the memory usage also.

http://www.ebi.ac.uk/~zerbino/oases/OasesManual.pdf

Hopefully it helps,

[jgibbons1]

I was trying to run velvet/oases with kmer from 25 toward 67 on my local HPC, setting it to run with 256GB of ram. Velveth/g performed well, while oases stopped after around 10 hrs and displayed an error.

job killed: mem 269731380kb exceeded limit 268435456kb

My command was oases transcripts_69 -min_trans_lgth 200...Any suggestion ?

Just out of curiosity, how many reads are you trying to assemble? Paired end or single end? Have you quality trimmed your data before assembling?