Trajectory analysis of combined single-cell and single-nuclei data

Hi all. I was wondering if I could get some advice around working with a combined sc and snRNA-seq dataset? I’m working with human tissue developmental data, with embryonic counts coming from sc and fetal from sn data. Obviously very interested in trying to identify trajectories, but not sure if it is viable given the source and type of biases. I’ve integrated and corrected my sc and sn data with scVI-scANVI, but since it doesn’t batch-correct the counts, I am not sure how “analyseable” this combined data is for trajectory stuff… What is the general approach here, if any? Should this data be analysed completely separately? I am specifically interested in the possibility of doing this analysis using scVelo, Palantir and CellRank.

Any thoughts would be highly appreciated.

Velocity using spliced and unspliced counts would’ve very weird in this data (sn just contains much more unspliced counts). Pseudotime approach can be performed in latent space as e.g. in Single-cell multiomic analysis of thymocyte development reveals drivers of CD4+ T cell and CD8+ T cell lineage commitment | Nature Immunology. Please tag future questions about scVI-tools with scVI-tools this makes it easier to see these posts.