The biggest difference here is that I'm formalizing the idea of a transmutation API between data structures as an Xarray accessor with functions dynamically added to the accessor based on the structure of the xr.Dataset currently being used (to have autocomplete + documentation reflect only what should be possible).
This issue should track any more requirements that pop up related to this, most likely from finding new types of analysis we want to support.
See https://github.com/pydata/xarray/issues/3959 for a discussion on how to approach this.
I refactored some old experiments from https://github.com/related-sciences/gwas-analysis/issues/15#issuecomment-605212471 into data_structures.ipynb (internals).
The biggest difference here is that I'm formalizing the idea of a transmutation API between data structures as an Xarray accessor with functions dynamically added to the accessor based on the structure of the xr.Dataset currently being used (to have autocomplete + documentation reflect only what should be possible).
This issue should track any more requirements that pop up related to this, most likely from finding new types of analysis we want to support.