scPhyloGRN: gene regulatory network inference guided by lineage topology in single-cell developmental data.

scPhyloGRN improves regulatory edge prioritization in single-cell developmental data across 17 dataset-reference settings spanning four mouse embryos and one Caenorhabditis elegans dataset.

• Why it matters: GRN inference from single-cell transcriptomes is hard because gene expression reflects both cell state and shared ancestry, and lineage-aware regulatory inference is still challenging.
• What they did: The supervised scPhyloGRN framework combines single-cell gene expression with cellular phylogeny to prioritize directed TF–target candidates, using lineage-guided graph construction, a dual-stream encoder with attention, and lineage-consistency regularization.
• The result: Compared with ten methods, scPhyloGRN shows an overall advantage in regulatory edge prioritization; removing lineage information lowers AUPRC, and branch- and regulator-level analyses improve ranking and recovery of reference-supported candidates.

AI summary of the abstract · may contain errors · not medical advice · read the paper.