Harness the Heterogeneity in Omics Studies
As high-throughput technologies being developed and widely applied in biomedical research, omics studies and datasets have been accumulated in public domain. Although these studies generated numerous biological findings, omics studies have been criticized for low reproducibility and inconsistent results due to high heterogeneity among studies addressing the same or similar biological questions. In this talk, I will discuss a Bayesian hierarchical model we recently developed to draw robust conclusion by combining multiple studies, while controlling for the heterogeneity among them. In addition, this method can also explore the potential biological meaning of the heterogeneity by a clustering analysis. I will demonstrate the usage of this method using real data examples. In the second part of my talk, I will discuss another method that we developed to accommodate the heterogeneity in effect sizes among different biomarkers when conducting SNV-set analysis in genome-wide association studies. I will discuss the application of this method in detecting dense and sparse signals in various studies.