Fig. 1: Overview of the scMORE.

a, Utilizing single-cell multimodal measurements to construct a global TF-GRN based on a GLM. b, A modified cosine similarity method is applied to infer the specificity score of each TF and its target genes (CTS) within a sub-GRN (eRegulon) in a specific cell type. c, Genetic association signals from GWAS summary statistics are mapped to peaks (PR), linked to corresponding genes (PS) and disease-specific genes (GS) are identified using MAGMA or FUMA gene-scoring methods. The GRS for each node is calculated as the product of GS, PR and PS, representing the genetic correlation of the node with the trait. d, The TRS is calculated for each eRegulon in a particular cell type by integrating the GRS and CTS scores. The s.d. term is included as a penalty to control the deviation between GWAS signals and gene expression. e, MC Model: 1,000 sets of matched control eRegulons are generated for each eRegulon in a given cell type. MC sampling is then used to calculate a P value for each eRegulon based on the empirical distribution. f, scMORE outputs. The outputs include (1) TRS, GRS and CTS values for each eRegulon associated with a specific disease in a given cell type; (2) MC-based empirical P values for each eRegulon within the specific cellular context.