Hey, I just want to share a tutorial on running GRN analysis and downstream AUCell umaps and do some self-advertisement on a tool we recently developed. I'm sharing it here because I saw some posts complaining about the speed issue that we think we have resolved.

Here is the link to the tutorial: https://tuftsbcb.github.io/RegDiffusion/downstream_with_pyscenic.html.

Basically, by replacing the grnboost step with regdiffusion, we can significantly reduce the computation time (hours/days -> minutes/seconds). The AUCell umap results are also much better compared with standard dimension reduction pipeline.

Check it out! Feel free to reach out with any questions.