Abstract / Summary
Abstract Background Ischemic stroke induces cell type specific transcriptional reprogramming, yet few computational tools resolve condition specific regulatory rewiring across cell types. Methods We developed StrokeCell-Reg, a Python toolkit that couples the stroke weighted prior database StrokeRegDB with ContextDiffReg. The module infers disease and control regulatory networks for each cell type and scores differential edges by bootstrap stability. We benchmarked the method on synthetic single cell data and applied it to mouse middle cerebral artery occlusion and human intracerebral hemorrhage datasets. Results On synthetic data, ContextDiffReg recovered planted rewiring with a mean AUPR of 0.60 ± 0.11 and AUROC of 0.85 ± 0.02. In a mouse MCAO atlas of 58,340 cells, it prioritized Nfkb1, Rela, Stat1, Stat3, Hif1a, Jun, Fos and Cebpb in microglia. Sixteen of the top 20 microglia regulators were shared across six mouse cohorts. A prior free pySCENIC run returned an IRF, IKZF and ETS myeloid program rather than NF kB, STAT or HIF regulons. Public PU.1 perturbation data supported the Spi1 target program in a directionally mixed manner, whereas a decoupled simulation suggested that ContextDiffReg can potentially recover rewired edges that target differential expression alone misses. Conclusions StrokeCell-Reg is a prior aware and cross dataset reproducible prioritization framework. Its comparatively robust output is the reproducible ranking of canonical stroke regulators. The de novo IRF, IKZF and ETS axis and the Spi1, PU.1 module are presented as testable hypotheses.