This project developed new models to estimate and assess the genotype frequencies of polyploid populations. Polyploids are organisms that have more than two sets of their genomes. Unlike humans, who are diploids (having two sets), many organisms are polyploids, including important agricultural crops like blueberries, potatoes, and strawberries. A crucial component in many agricultural and evolutionary studies involving polyploids is the variation of DNA from one organism to another. For instance, at a locus on the genome of a tetraploid (which has four sets), some individuals might have four adenine (A) copies and zero cytosine (C) copies, some might have three A's and one C, others two A's and two C's, and so forth. The proportions of individuals with each DNA composition are known as genotype frequencies.

This project yielded three publications that developed methods to test various assumptions about these genotype frequencies. This project devised tests to determine whether a population is in Hardy-Weinberg equilibrium, a common assumption in many evolutionary studies. This project created tests to check if a population is panmictic (randomly mating), typical in many natural populations. This project also developed tests to assess if a population adheres to the law of Mendelian segregation (tests for segregation distortion), a frequent query in agricultural breeding experiments. These are all tests of the assumptions of the genotype frequencies. The publications also presented strategies to estimate genotype frequencies and related population genetic parameters based on different assumptions about their forms. A fourth publication clarified some confusions in the literature regarding genotype frequencies.

This project developed novel strategies to implement our testing procedures. Our tests employ a novel approach based on constructing a statistic, called a U-statistic, which is zero on average. This statistic is a function of the genotype frequencies under various assumptions, such as the polyploids' meiosis form. Since polyploids, depending on their evolutionary history, have different meiosis models, this project developed a novel meiosis model in tetraploids, linking different polyploid types. Our U-statistic approach is distinctive in population genetics, where many methods adopt a likelihood approach based on a generative model for observed genomes. This project also explored using this U-statistic method to study population structure, which examines how some genomes in a population are more similar than others.

Our methods have broader applications to agriculture and evolution. Many plant breeding programs aimed at the agricultural improvement of polyploid crops use genomic approaches to identify locations on the genome associated with desired traits, such as yield and hardiness. Numerous evolutionary genomic studies involve polyploid organisms, as all modern angiosperms (flowering plants) evolved from ancient polyploids. However, genomic data can be complex and prone to corruption by aspects of the biological assays designed to measure it. Our tests can serve as a quality control measure to identify problematic genome locations resulting from data corruption. The Hardy-Weinberg equilibrium tests can verify if many evolutionary methods' assumptions are valid. The random mating tests can highlight issues at certain genome locations in natural populations, often used in evolutionary studies, where random mating frequently occurs. The segregation distortion tests can assist breeding programs in identifying locations not following Mendelian segregation. Therefore, our tests can be integrated into quality control procedures to enhance evolutionary and agricultural studies, potentially advancing our understanding of plant evolution and improving crop quality. The easy-to-use software developed during this project, the hwep and menbayes R packages, facilitates access to our tests for scientists and plant breeders.

This project also funded educational initiatives, producing publicly available lecture materials for two courses. One course focused on advanced programming using the R statistical language, emphasizing genomic analysis methods in R. The other course covered introductory statistical genetic analysis at the graduate and upper undergraduate levels. Twelve student researchers funded by this project learned the fundamentals of statistical genetics and computational biology, with two writing master's theses on polyploid genetics.

Last Modified: 03/28/2024
Modified by: David Gerard