Allow del adata.X, adata.X = None

[rcannood]

Feature proposal for #464.

Need to test:

• Getter (test_base.py > test_x_is_none())
• Setter (test_base.py > test_x_is_none())
• Deleter (test_base.py > test_x_is_none())
• Initializer (test_base.py > test_x_is_none())
• Views
  • Check that .X = None doesn't change other view behavior
  • a[idx].X doesn't do anything weird
  • copying views still works
• _subset_inplace
• Copying
• Backed mode
• Concatenation
• Transpose
• Methods which default to values from X, e.g. obs_vector, to_df.
• IO (seems to work, needs testing)
  • Maybe specialized writers like to_csvs will need some attention here?
• Raw (pretty similar set of things to be checked)
  • Initializer
  • Backed
  • Copies
  • Concatenation
  • IO

[ivirshup]

I don't think we currently allow X to be None (#331).

I have come across a case recently where this would be useful, though I think it would take some doing to make the entire codebase work again if we allowed this. X being an array with shape is pretty baked into the data model at this point.

Maybe this could be opened as a feature request issue, and you could pitch your use case?

[rcannood]

Thanks for your input.

X being None is already part of the unit tests: test_base.py#L46-L47. Also, if X would not be allowed to be None, why is there a check in the code I changed? ;)

A possible use case I have is when working with read-only h5ad files: if your input is immutable, it may be interesting to exclude X from the output file to reduce the file size.

[ivirshup]

At one point it was allowed, but I can say for certain that a lot of code has been written since then that does not expect X to be None.

For example:

a = ad.AnnData(X=None, shape=(100, 200))

a.copy()

Traceback

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-12-02bb7baf10a6> in <module>
----> 1 a.copy()

~/github/anndata/anndata/_core/anndata.py in copy(self, filename)
   1428                 X = _subset(self._adata_ref.X, (self._oidx, self._vidx)).copy()
   1429             else:
-> 1430                 X = self.X.copy()
   1431             # TODO: Figure out what case this is:
   1432             if X is not None:

AttributeError: 'NoneType' object has no attribute 'copy'

At one point I think I saw _X being None as an indicator for if the object was in backed mode.

If this is something we were to support more, it would need a fair amount of testing.

[rcannood]

Alright :) I hereby turn this PR into a feature request. I'd like X to be optional. I mean, it already is, but apparently more thorough testing is needed to verify that it works properly.

What is the opinion of the major anndata developers? 👍 or 👎?

If we agree X should be able to be None, we can use this PR to add tests and fix the code where necessary.

[ivirshup]

Oops, I linked the wrong issue before. #330 is where this has come up before. Also I'm going to open this discussion as a new issue so it's more visible.

[rcannood]

Plus it's much easier to review and merge small PRs. Let's keep that PR's scope to just fixing up the .X property.

@ivirshup Alright, shall we merge this PR in order to resolve the first box in #467?

[ivirshup]

Great, thanks for implementing this and making the suggestion!