The `CreatePairedObject` function contains a logic error that leads to mismatched `counts` matrices between ATAC and RNA modalities.

[noralis-bit]

When generating paired RNA and ATAC counts datasets based on the df.pair information, the Initial code did not correctly reorder the data according to the cell IDs provided by df.pair, which led to mismatches and disordered pairing.

Initial Code

rna.counts <- LayerData(obj.rna, assay = rna.assay, layer = "counts", cells = df.pair$RNA)
atac.counts <- LayerData(obj.atac, assay = atac.assay, layer = "counts", cells = df.pair$ATAC)

✅ Proposed Fix

To ensure the output matrices are strictly ordered by the pairing information in df.pair, we should reindex the columns after extraction:

rna.counts <- LayerData(obj.rna, assay = rna.assay, layer = "counts", cells = df.pair$RNA)[ , df.pair$RNA]
atac.counts <- LayerData(obj.atac, assay = atac.assay, layer = "counts", cells = df.pair$ATAC)[ , df.pair$ATAC]

@lzj1769

[lzj1769]

Hi @nh-codem

Thanks for pointing this out. Can you make a pull request?