Awarded in 2026
Home Department: Genetics
Faculty Advisor: Jonathan Pritchard (Genetics and Biology)

Title: Biobank-Scale Discovery of Pathogenic Mutations via Multi-Omics Integration

Abstract:
Identifying pathogenic mutations that impact human health has been a central challenge in human genetics, particularly because many such mutations are under strong negative selection and therefore remain at ultra-low frequencies. The recent emergence of biobank-scale whole-genome sequencing provides unprecedented opportunities to systematically detect these mutations, which would be otherwise unseen in moderate-size sequencing cohorts. However, classical population genetics approaches are limited in scope to handle biobank-scale datasets, and they tend to be poorly integrated with the rich functional genomic annotations, which also provide parallel evidence of the mutation effects. 

This project aims to develop a unified Bayesian framework to estimate single-bp level mutation-specific selective constraint by integrating population genetic theory with multi-omics data. Existing methods largely focus on loss-of-function mutations and rely heavily on allele frequency as a proxy for selection, overlooking the majority of variants—such as missense and noncoding mutations—and failing to incorporate diverse functional annotations (both experimental and predictive). To address these limitations, we propose a model that combines allele frequency likelihoods derived from population genetics with a deep learning–based prior over selection constraints informed by functional features, including gene expression (GTEx), chromatin state (ENCODE, chromBPNet), protein structure (AlphaFold2), and variant effect predictions (e.g., SpliceAI, AlphaMissense). 

Leveraging large-scale datasets such as All of Us and the UK Biobank, we will infer selection constraint genome-wide and prioritize mutations with high pathogenic potential. Model performance will be evaluated using ClinVar annotations and rare disease cohorts. By integrating population genetics theory with modern biobanks and functional genomics data, this work aims to improve the discovery and interpretation of pathogenic mutations at an unprecedented scale and resolution.