[Feature Request] Save PyCytoData Metadata

[kevin931]

What feature would you like to propose?

We should think about an easy interface to save and load the metadata in PyCytoData, especially the cell types and sample indices. Currently, only channel names and the expression matrices are saved, but if users performed cell-typing or they just wish to export a built-in dataset or import their own, we need to allow the full usage of PyCytoData.

Is your feature request related to a current problem? Please describe.
No.

How do you envision this feature to be implemented?

I think the new feature should satisfy the follow criteria:

• It is plain-text based. We should not resort to fcs to do this.
• No additional external dependencies. On the surface, pandas may sound like a good idea, but this is debatable and needs further consideration.
• The saved metadata should have reasonable extensibility. We don't want to bloat the directory in the future, in case new metadata are added for future support.

This can live in the FileIO module or be part of the PyCytoData. If it allows the recontruction of a PyCytoData object, then the latter makes a lot sense.

Are there any existing alternatives or workarounds?
Technically yes, we can save metadata separately using any NumPy or python methods. But we don't have any documented or supported way of doing this.