I did this & the # of genes changed the clusters but this is ++compute intensive. Anyone know of any good reviews now on params used for cell type clustering?

The cell type clustered affects any claim of this sort, I find that interesting, would be nice to know if anyone has written about it.
This is from the Illumina #AGBT25 workshop about their ST tech.

But, from our observations, under heavy lateral diffusion (real or simulated), clustering on large sets of genes (i.e. “whole transcriptome”) leads to niche- rather than cell-type clustering.

This might be problematic.

Comments