Add gene cycle scores to cells?

[jashapiro]

Proposed analysis

It might be useful for all cells that we have to be classified with some scores of their cell cycle state and/or scores for markers of proliferation generally.

Scientific goals

For many cancer types, we might be able to improve classification of tumor types if we were to add cell cycle scores to the cells. Ideally, we would be able to differentiate proliferative cells from non-proliferative. The goal at this point is not to try to regress out any of the cell cycle effects, but to evaluate if and how they might be useful to aid classification in general, particularly for the identification of malignant cells.

Methods or approach

I have not yet had a chance to look into this in much detail. I know that there are functions such as scran::cyclone and Seurat::CellCycleScoring() which will allow classification into G1, G2/M and S phases, most commonly, but these do not readily allow distinction of G0 cells, which would be ideal.

I found at least one classifier that attempts to include G0 specifically (at least for neuronal cells): ccAFv2 (O'Connor et al. 2025), but I have not evaluated the literature extensively. We would presumably want to do more of that before starting on implementation!

Relatedly, I am also curious about the performance of scMalignantFinder (Yu et al. 2025), but that might be a different analysis.

Existing modules

Not that I am aware of currently, but I can imagine that we would add some that could depend on these results.

Input data

Process gene expression matrices, either SCE or AnnData, depending on the tool.

Scientific literature

• Cell Cycle assignment in OSCA
• Seurat cell cycle scoring
• O'Connor et al. 2025
• Yu et al. 2025

Other details

While I would expect implementing these scoring methods will be fairly fast, the harder challenge will be evaluating their utility.