Inquiry Regarding DRVI Model Compatibility with scANVI for Embedding Optimization?

I am writing to inquire about the compatibility of the DRVI model with scANVI for embedding optimization. Specifically, I would like to know:

  1. Whether the DRVI model supports using scANVI to optimize embeddings, and if so, how this integration can be implemented.

  2. Whether this approach is recommended (e.g., for improving latent representations, scalability, or downstream tasks like clustering or visualization).

I have reviewed the DRVI documentation and codebase but could not find explicit information on this combination.

Thank you for your time and contributions to the community. I look forward to your guidance.

Hi, thanks for your interest. These are two distinct models and DrVI is not part of scvi-tools. So they are not compatible with each other and reaching out to the DrVI developers directly on their GitHub, might actually help you to understand whether they are planning for a (semi-)supervised model in the future.