Hello all,
I have decided to integrate my single cell dataset using rPCA rather than CCA because I thought using CCA overcorrect my dataset. I know there is two approaches using rPCA: to use references in "FindIntegrationAnchors" function and the reason is that it reduces the number of pairwise comparisons and is less computationally intensive which is good for large dataset. My first question is that as using references limits finding anchors by reducing the number of pairwise comparisons do I miss anything doing this method? My second question is that if I want to use rPCA and my dataset is not that large (to make me set references) do you recommend to do this method using references or not? I mean is the only purpose of setting reference to make it less computationally intensive? Any comment is greatly appreciated
Paria
