Possible issue with very disparate numbers of cells #490
Description
Brian mentioned that RISE has some weird behavior, possibly when the number of cells per condition is very disparate. He took perturb-seq data and had collected all the control cells into one condition. This caused the condition to have a universally strong weight, even after the correction step. Splitting these cells across conditions didn't have this problem, though. There's public perturb-seq data where we could examine this.