Oases/Velvet error memory

[JManuelMonroyKuhn]

Hello!

I was running Oases/Velvet with this command

python /home/hpc/uj502/di29kiz/oases_0.2.8/scripts/oases_pipeline.py -m 35 -M 45 -s 2 -o '/naslx/projects/uj502/di29kiz/oases/LA2750' -d '-fastq -shortPaired -separate /naslx/projects/uj502/di29kiz/libraries/NG-7404_LA2750tpool1p.fastq /naslx/projects/uj502/di29kiz/libraries/NG-7404_LA2750tpool2p.fastq' -p '-ins_length 229'

but I got this error

velvetg: Can't calloc 18446744071712318997 Annotations totalling 18446744041753829712 bytes: Cannot allocate memory
Traceback (most recent call last):
File "/home/hpc/uj502/di29kiz/oases_0.2.8/scripts/oases_pipeline.py", line 121, in <module>
main()
File "/home/hpc/uj502/di29kiz/oases_0.2.8/scripts/oases_pipeline.py", line 115, in main
singleKAssemblies(options)
File "/home/hpc/uj502/di29kiz/oases_0.2.8/scripts/oases_pipeline.py", line 43, in singleKAssemblies
assert p.returncode == 0, "Velvetg failed at k = %i\n%s" % (k, output[0])
AssertionError: Velvetg failed at k = 35
[0.000001] Reading roadmap file /naslx/projects/uj502/di29kiz/oases/LA2750_35/Roadmaps

I have around 350 million paired reads. I'm working on a cluster that uses SLURM. Any recommendations? What should I do?

[dpryan]

You're going to want to work with fewer reads than that...your cluster is trying to use 18 exabytes of memory, which it certainly doesn't have. Give digital normalization a try, for example with the tools from Titus Brown's group. That should heavily reduce the total number of reads to assemble and increase the overall assembly quality. Also, if you haven't already done so, make sure to quality/adapter trim the dataset.