Abstract
Mapping enhancers and target genes in disease-related cell types provides critical insights into the functional mechanisms of genome-wide association studies (GWAS) variants. Single-cell multimodal data, which measure gene expression and chromatin accessibility in the same cells, enable the cell-type-specific inference of enhancer-gene pairs. However, this task is challenged by high data sparsity, sequencing depth variation, and the computational burden of analyzing a large number of pairs. We introduce scMultiMap, a statistical method that infers enhancer-gene association from sparse multimodal counts using a joint latent-variable model. It adjusts for technical confounding, permits fast moment-based estimation and provides analytically derived p-values. In blood and brain data, scMultiMap shows appropriate type I error control, high statistical power, and computational efficiency (1% of existing methods). When applied to Alzheimer's disease (AD) data, scMultiMap gives the highest heritability enrichment in microglia and reveals insights into the regulatory mechanisms of AD GWAS variants.