Error in ATAC single-cell workflow

[jiyuan426-tech]

Please make sure these conditions are met

• I have checked that this issue has not already been reported.
• I have confirmed this bug exists on the latest version of omicverse.
• (optional) I have confirmed this bug exists on the main branch of omicverse.

What happened?

When I try to use ov.epi.pp.add_tile_matrix(adata, bin_size=500) for ATAC analysis, an error occurs

Minimal code sample

ov.epi.pp.add_tile_matrix()

Error output

building tile matrix: 568 cells x 6,176,584 tiles (500 bp bins, strategy-paired-insertion)
ExceptionValueError
Data matrix has wrong shape (568, 6176584), need to be (568, 0).

Versions

ExceptionValueError
Data matrix has wrong shape (568, 6176584), need to be (568, 0).

[Starlitnightly]

The bug is in the backing epione package that ov.epi.pp.add_tile_matrix delegates to: add_tile_matrix assigned the (n_cells, n_tiles) tile matrix to adata.X before replacing var, but the AnnData from import_fragments has 0 vars, so AnnData rejected the new X (Data matrix has wrong shape (568, 6176584), need to be (568, 0)).

Fix (rebuild X + var together in place): aristoteleo/epione#42. Once that's merged and released, pip install -U epione resolves it — no omicverse change needed. Verified on a minimal 0-var AnnData + tiny fragments file: builds the tile matrix correctly, mutates in place, preserves uns.

[mengxu98]

The error has been traced to epione, which implements ov.epi.pp.add_tile_matrix: updating the matrix and variable metadata separately causes a shape mismatch when the input AnnData has no variables.

A fix has been submitted in aristoteleo/epione#42 and is still awaiting merge. Please follow that PR for progress. Once the fix is merged and released, an epione version containing the fix will be required.

We are closing this OmicVerse issue because the fix is being tracked in the epione repository.