Scientific Interests

• Statistical methods and machine learning approaches for detection of tumor-derived DNA based on genome-wide cell-free DNA fragmentation patterns. Applications include the early detection of cancer and monitoring disease in late stage cancer patients.

• Methods and software for modeling technical and biological variation within and between large-scale studies with high-throughput data. Applications include Bayesian hierarchical models for differential gene expression analysis in multiple studies and Bayesian hierarchical models for copy number estimation.

• Modeling the contribution of molecular characteristics to cancer risk using models that incorporate measurement error of genomic variants.

• Bayesian approaches to model uncertainty in evolutionary models of cancer progression.

• Statistical models for co-occurrence of somatic sequence variants, copy number variants, and rearrangements in large-scale databases.