Research
My dissertation develops statistical methods for rare variant association tests and variant prioritization in biobank-scale data.
VACANT: annotation-stratified rare variant association analysis
VACANT uses continuous functional annotation scores to stratify rare variants, tests their association with a trait adaptively, and estimates effect sizes for each annotation stratum. I presented it as a platform talk at IGES 2026, and earlier as posters at ASHG 2024, ASHG 2025, and the 2025 Taipei International Statistical Symposium and ICSA International Conference.
VACANT-M: accounting for relatedness and population structure
VACANT-M is a mixed-model extension of VACANT, so that the test accounts for genetic relatedness and population structure in biobank cohorts (poster at IGES 2025).
SHARC: recalibrating coding variant annotation scores
SHARC improves severity prediction for loss-of-function SNVs and indels, and recalibrates coding variant annotation scores across variant types for clinical variant classification (poster at ASHG 2026).
Earlier work
- Causal mediation analysis with causally ordered mediators under the Cox proportional hazards model (Statistics in Medicine, 2019), extended to a semiparametric approach for semi-competing risks.
- Adaptive information borrowing for the Cox model to improve prognostic prediction in triple-negative breast cancer.
- A large-scale genome-wide association study of stuttering, to which I contributed a secondary heterogeneity analysis (Nature Genetics, 2025).
Papers and manuscripts are listed on the Publications page.