A paper has recently been making the rounds that suggests PCA outperforms PEER on RNA-seq data. The paper is here: https://www.biorxiv.org/content/10.1101/2022.03.09.483661v1.full.pdf or here: https://genomebiology.biomedcentral.com/articles/10.1186/s13059-022-02761-4 The twitter discussion is here: https://twitter.com/jsb_ucla/status/1580023606721269760?cxt=HHwWgMDS9arGr-0rAAAA

It seems like a careful study, but I can't get it out of my mind that I thought PEER performed better in tests I'd done myself in the past (but I don't have access to those simulations anymore so maybe I'm misremembering). My impression is that they didn't use real RNA-seq data in their simulations, so I wonder if the real sources of batch effects and bias are more complicated than what they simulate, in which case PCA may perform worse.

Wondering if anyone else has a hot take on this.