What could be a better K-mer for my WGS assembly task using ABySS, with my current hardware profile.

Question

[Vijith]

Performing a whole genome sequencing project for a plant genome, with an estimated genome size of 2 GB. I have 2*150 bp, Illumina Novaseq X Plus 10B (125M/Lane) reads. I am using AbySS 2.0 assembler for the assembly task, for which the AbySS was compiled from source. Previously, I had issues with the installation of ABySS, and it was addressed at "C++ compiler cannot create executables", while installing ABySS 2.0 assembler.

I am using an IBM server with a hardware capacity of 244 GB DDR3 RAM, and 8 cores from 2 processors.

I would like to know what can be an ideal K-mer size that I could pass to the run command, so that the process doesn’t require more memory resources than I currently have. I know there is no clear-cut answer to this, but I was thinking if K-mer value of 32 would be so low that it would make the assembly process run for an extended period.

NB: Kindly let me know if you need any additional information.

Answer 1

[shelkmike]

ABySS can use Bloom filters to reduce RAM usage. If you use Bloom filters, I'm sure that 244 GB RAM will be enough.
The optimal k-mer size (optimal for the best assembly, not for minimal RAM utilization) can be determined by Kmergenie (http://kmergenie.bx.psu.edu/)