Thank you for developing such a powerful tool for RNA velocity analysis. I’m currently working through the VeloVAE tutorial and exploring the scripts published alongside the paper. However, I’m facing some difficulties understanding certain aspects:
Differences Between VeloVAE and VeloVAE (Full VB):
Could you clarify the core differences between training a VeloVAE model and a VeloVAE (Full VB) model? I noticed in the tutorial that “The full VB model is a slight extension of VeloVAE by treating the rate parameters as random variables.” Could you expand on this distinction, specifically should both models be trained? I’ve read your paper, you could also point me to the relevant section that addresses this.
Usage of Rate Prior:
In the sample tutorial for VeloVAE, I noticed that a rate prior was applied. However, in the scripts related to your published paper, this rate prior was not included. Could you explain more on that too?
Hi VeloVAE team,
Thank you for developing such a powerful tool for RNA velocity analysis. I’m currently working through the VeloVAE tutorial and exploring the scripts published alongside the paper. However, I’m facing some difficulties understanding certain aspects:
Could you clarify the core differences between training a VeloVAE model and a VeloVAE (Full VB) model? I noticed in the tutorial that “The full VB model is a slight extension of VeloVAE by treating the rate parameters as random variables.” Could you expand on this distinction, specifically should both models be trained? I’ve read your paper, you could also point me to the relevant section that addresses this.
In the sample tutorial for VeloVAE, I noticed that a rate prior was applied. However, in the scripts related to your published paper, this rate prior was not included. Could you explain more on that too?
Thank you in advance for your help!