Scalable Population Genetic and Phylogenetic Inference Using Large Samples of Microbial Data

This project will enable us to more effectively use large amounts of DNA and other genomic data to study human history, natural selection, pathogen evolution, and other topics that are important to science and human well-being. Examples of the types of questions it addresses include: When did humans migrate out of Africa? How did polar bears fare during the last global warming event? Does being tall confer evolutionary advantages? What is the current status of the COVID-19 pandemic, and when will it end? Although definitively answering these questions is challenging, evolution furnishes clues about them in the form of genetic variation. These clues can be decoded using mathematical models to analyze DNA samples from current populations. The amount of available genetic data has increased dramatically in recent years, and consequently faster and more accurate analytical methods are needed to fully utilize these rich new sources of information. This project will develop those methods. In addition, it will facilitate the creation of new curriculum materials designed to educate students about quantitative genetics and computational biology.

This project will develop new and scalable methods for phylogenetic and population genetic inference, with a particular focus on analyzing pathogen genetic data. Although these areas are already quite mature, historically many methods designed to analyze genetic data made modeling assumptions based on human biology and data availability. Such assumptions complicate efforts to study the genetics of species whose biology is very different from humans, even though these species can have important impacts on human health. This project addresses these shortcomings through: the creation of novel methods for phylogenetic inference which can adapt to the amount of phylogenetic signal present in the data; faster likelihood-based phylogenetic network inference methods which allow for horizontal gene transfer or other reticulate events; variational methods for rapidly inferring epidemiological parameters from pandemic data; and new applications of coalescent hidden Markov models which are faster and have less bias than existing methods.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.