MASH in preprocessed reads to choose coassmbly groups:

[EisenRa]

Been thinking about our recent coassemblies with the pyranese lizards (which showed that coassembling with samples that are too different reduced preformance). I came across this idea today when skimming the EMP500 protocol (https://github.com/biocore/emp/blob/master/methods/methods_release2.md):

Running pairwise MASH on the metagenomic reads of all samples, then selecting samples for coassembly based on a minimum MASH distance.
Pretty nifty idea, which could give us a more quantifiable justification for which samples we choose to coassemble together.

They choose MASH distances < 0.1 for coassembly, but this could be explored a bit more.