A colleague of mine and I are wanting to perform co-assemblies of soil metagenome datasets. We can do this using Megahit but would like to try using SPAdes 3.13 as comparative studies have shown that this produces longer contigs and an overall higher quality assembly. Unfortunately SPAdes runs out of memory on our current server setup (Gb RAM 374 Gb). We can’t run the assembly across nodes as SPAdes can only run on one node but people in the US have been able to assemble large soil metagenomes with 500 Gb+ RAM. Would it be possible to set up a single group VM for the MPEG group solely for the purpose of running these large co-assemblies? Jobs of this size would be run fairly infrequently but would probably last a few days each time.
We are unable to provide a permanent instance of that size. However, we are able to provide large memory instances on a temporary basis. Would that suit your group?
All the best,