scvi-tools
scvi-tools copied to clipboard
Finalize scBasset
- [x] Verify that all parameters are being initialized equivalently, alter the ones used here if needed
- [x] Add more content to the tutorial that reproduces their results, allowing us to have some form of reproducibility
- [ ] Complete the user guide
@adamgayoso Thoughts on adding a batch correction portion to the tutorial using the buenrostro dataset from the original manuscript?
I think we should and the maybe use scib metrics on it
I'm wondering whether we should add an option in setup_anndata to indicate whether the data is transposed already or not, where if it's not transposed, we do it on our end. It might be confusing for users when you have to bdata = adata.transpose() but then the latent space is saved back into the original adata.
I agree this needs improvement. I'm a fan of what you see is what you get and it would be confusing if model.adata was transposed