I will talk about a few statistical and computational methods we have developed over the past few years to give you a flavor of the type of work we do in our group. My talk will focus on two distinct application areas: genome-wide association studies and spatial multi-omics studies. Specifically, I will talk about Dirichlet process regression, or DPR, a non-parametric Bayesian regression method that flexibly and adaptively models the effect size distribution to enable accurate and robust polygenic risk prediction across a broad spectrum of genetic architectures. I will talk about SPARK, a method that allows for rigorous statistical analysis of spatial expression patterns in spatial transcriptomics, along with its non-parametric extension, SPARK-X, for scalable detection of spatially expressed genes in large spatial transcriptomic studies. If time allows, I will also talk about a scalable multi-ancestry variational fine-mapping method, MESuSiE, that accounts for the diverse linkage disequilibrium pattern observed in different ancestries while explicitly modeling both shared and ancestry-specific causal SNPs; as well as a spatially informed cell type deconvolution method, CARD, that leverages cell type specific expression information from single cell RNA sequencing for the deconvolution of spatial transcriptomics.