Haplotype-based inference of hidden structure in the human genome

The rapid emergence of high-throughput genomic technologies—such as low-coverage sequencing (LCS), ancient DNA (aDNA), and long-read sequencing (LRS)—offers unprecedented opportunities to explore human evolutionary history and advance medical genetics. However, the full potential of these data remains underexploited due to analytical and methodological limitations: for example, low coverage aDNA (<0.5x) is typically excluded ancestral recombination graph (ARG) analyses, while low structural variant (SV) genotyping accuracy and the lack of reliable imputation panels limit population-scale association studies. This project proposes the development of two innovative haplotype-based statistical methods to overcome these barriers and unlock novel biological insights. First, I will establish a unified framework that integrates ARG construction with genotype imputation, enabling robust large-scale genealogical and demographic analyses of both modern and ancient low-coverage genomes. Second, I will develop a probabilistic haplotype-based approach for structural variant (SV) genotyping and imputation with long-read sequencing data, applying it to large population cohorts. These methods will be validated on cutting-edge resources, including the UK Biobank and the Allen Ancient DNA Resource, and disseminated through open-source software.By bridging methodological gaps in aDNA and SV analyses, this project will deliver tools for demography inference using ARGs, SV-informed trait mapping, and the integration of diverse genomic data in population and medical genomics. The first application of the tools will have profound impacts. The unlocked high-resolution human history will contribute to public discourse on the definition and biomedical use of genetic ancestry. Discoveries of disease-associated SVs and underlying biological mechanisms will inform drug development, personalized treatment, and supporting industrial innovation.