general question embedding projection #23
Description
I generated my embedding using Seurat, utilizing the raw count matrix rather than the spliced matrix. Following your instructions to project velocities onto this UMAP embedding, the process involves recalculating new PCs and neighbors, using different genes compared to those initially employed to create the embedding. is this still valid?
Additionally, I have data from 8 distinct donors/batches. How should I handle this aspect in the analysis?