Should AnnData be incorporated into (or even replace) the MultiOmics Expression data?

[JonnyTran]

As you know, the hugely popular AnnData library initially developed for Scanpy has been generally easy to use and applicable to a wide-range of bioinformatics data analysis. The library is quite mature and offers a lot of data munging capability, downstream analysis with scanpy, compatibility with the various IO file types. It has a lot functions (and more) that overlap with the MultiOmics Expression data, such as:

• Handling tabular data where rows correspond to samples (or observations) and columns correspond to genes (or variables).
• Coupling of observation annotations and variable annotations using dataframes.

I think it can greatly serve the users (bioinformaticians) if the backend of Expression data would be an AnnData type, or at least able to export to an AnnData object.

What are your thoughts on this? @gawbul