Principled phylogenomic analysis without gene tree estimation

This project aims to improve the estimation of species trees from genomic datasets. This estimation is challenging because different genomic regions evolve under processes that make their evolutionary histories (i.e., gene trees) discordant. This issue is exacerbated by widespread gene tree estimation errors in modern phylogenomic analyses. To address this challenge, this project's primary objective is to devise innovative mathematical, statistical, and computational techniques to analyze phylogenomic datasets without relying on gene tree estimation. This approach will produce more reliable species tree estimates in the presence of confounding processes. Species trees provide an evolutionary and comparative context in which many biological questions can be addressed. They play a vital role in understanding gene evolution, estimating divergence dates, detecting adaptation, studying trait evolution, etc. The developed methods will enhance the precision of biological discoveries based on species trees, advancing research that utilizes phylogenies. The project includes interdisciplinary research training for graduate students as well as the involvement of undergraduate students recruited through local initiatives. New course materials based on the proposed research will be developed for existing graduate courses and be made available through the PI’s website. The project will leverage connections to NSF-funded interdisciplinary institutes. It is well established that different regions of a genome can evolve under different gene trees, due to processes such as incomplete lineage sorting, gene duplication and loss, and lateral gene transfer, complicating the estimation of species trees. Many methods that first estimate gene trees and then combine this information to estimate a species tree are known to have good theoretical guarantees, under the assumption that the true gene trees are known. That assumption is not satisfied in practice. Accounting theoretically for gene tree estimation error has proved challenging and few results are available. Building on prior work by the PI on the rigorous study of stochastic processes arising in this phylogenomic context, the proposed research will establish much-needed theoretical foundations for the analysis of multi-locus, multi-site datasets and the estimation of species trees without gene trees, including the development of novel estimators, the derivation of impossibility results and matching finite sample bounds, and the investigation of the effect of intra-locus recombination. This project will also enable the development of statistically rigorous, scalable algorithms. This interdisciplinary research will involve a close integration of applied probability, statistical theory, graph algorithms, and evolutionary biology. This proposal is jointly funded by the Mathematical Biology and Statistics Programs at the Division of Mathematical Sciences. 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.