Advice on creating metaCells

[jdarmitage]

Hi,

When it comes to clustering and generating metaCells from clusters, is there a rationale to determine how granular these clusters can be prior to network construction?

i.e. Are the ARACNe networks more robust when created from total CD8+ T cells or can these be performed on subsets of CD8+ T cells, to create separate networks for Naive CD8, Memory CD8, Effector CD8, Exhausted CD8 (etc.)

Secondly, if more cells are available can ARACNe networks be created from >250 metaCells with better accuracy? Will the results be massively different if running 4 clusters @ 250 metaCells (4 networks) versus 1 merged cluster, assuming equal numbers from each @ 1000 metaCells (1 network) for instance?

Best regards,
Jesse