Molecular population genetics of cis-regulatory variation

My dissertation has focused on elucidating the roles that natural selection, genetic drift, mutation, and demography play in shaping cis-regulatory variation. Using the framework of modern population genetics, I present computational, statistical, and empirical approaches that provide additional evidence for the importance of changes in regulatory DNA.; Chapter 1 is an in-depth introduction to the previous work done on the population genetics of cis-regulatory sequences and the methods used to infer the action of natural selection. I review a growing body of literature that has demonstrated many modes of selection, represented here by examples in chapters 2, 3, and 4. In Chapter 2, I present a statistical test of the neutral hypothesis created with Mark Rausher, and, using Cliff Cunningham's experimentally-evolved virus, show that this test is useful in distinguishing both between selection and demography, and among different selective hypotheses. Chapter 1 applies this test to previously published cis-regulatory sequences to further demonstrate its use. In Chapter 3 I demonstrate a novel form of selection that can only be detected at the level of the whole genome: selection against the constant influx of new transcription factor binding sites by mutation. Jason Stajich and Greg Wray collaborated on the informatics and on the creation and development of the idea. I further present a model of mine that is able to show that this selection is relatively weak. Chapter 4 presents an empirical study of cis-regulatory variants at a single locus in humans. I show that natural selection distinguishes among these variants and has different affects in different human populations. Nicole Soranzo and David Goldstein provided the neutral marker data, Matthew Rockman the non-human primate data. The F7 genotyping and all analyses and writing were done by me. Chapter 5 presents an empirical study of regulatory change responsible for species differences. I show that the difference between white- and purple-flowered morning glories is likely due to mutation at a single transcription factor locus. Mark Rausher collaborated in the creation of the project and on the expression assays. Chapter 6 presents a model of binding site evolution that builds upon the results found in Chapter 3. Analytical solutions and simulation studies define allele frequency distributions within populations, and lead to a statistical test of the neutrality of spurious binding sites. I performed all of the analyses and simulations, and Mark Rausher was integral to the creation and development of the model. (Abstract shortened by UMI.)...