Thesis (Ph. D.)--University of Missouri--Columbia, 2002. / Typescript. Vita. Includes bibliographical references (leaves 205-223). Also available on the Internet.

Some of life’s most important behaviors are guided by the sense of smell. Detecting and interpreting odor information influences food-seeking, predator avoidance, sociality, competition, mating rituals, and more, shaping how organisms interact with their environment. In vertebrates, odors are detected by olfactory sensory neurons (OSNs) of the main olfactory epithelium (MOE). OSNs rely on olfactory receptors (ORs) to recognize odorants and trigger neural activation. The OR gene pool is typically vast, containing between 200-4000 olfactory receptor genes across mammals, yet mature OSNs stably express only one gene from one allele. Data from mice show that ORs are anatomically restricted to designated sections of the MOE, but within these zones, OR expression appears mosaic and random. Since the discovery of the OR gene pool 30 years ago, deciphering how OSNs choose which OR they are going to express remains a central question. While multiple differentiation-dependent alterations to the OSN nucleus are required for OR expression, the most notable contribution comes from the organization of OR-gene specific enhancers, called Greek Islands (GIs), around the chosen allele. GIs use the transcription factors Lhx2 and Ebf1, as well as the coactivator Ldb1, to form a nucleoprotein complex known as the Greek Island Hub (GIH) to associate with the active OR gene and support its transcription. Bulk Hi-C data show that GIs form strong, specific, and singular associations with the active OR gene, suggesting a possible role for the GIH in singular OR choice. However, single-cell Hi-C analysis shows that multiple GIHs exist in every OSN with no clear differences between them, complicating the contribution of the GIH. Furthermore, ectopic OR gene activation is sufficient to drive association of an OR locus with a GIH and bias choice, suggesting a role for OR transcription itself in supporting its own stable expression. To clarify the genomic transformations that result in the formation of multiple GIHs, I performed combined scRNA-seq and scATAC-seq in the MOE. I determined that a selective inactivation event was taking place during the INP3-to-iOSN transition, where OSNs would silence a large fraction of the GI pool. GI inactivation takes place during a phase preceding OR choice, where OR expression is polygenic but skewed towards one OR. My single-cell Hi-C analysis verifies the presence of multiple GIHs per cell, with similar GI-GI interaction properties, but I also observe that the single active GIH contains much more specific GI-OR gene interactions than those in inactive GIHs. These architectural differences are supported by Liquid Hi-C and H3K27ac HiChIP analysis where I observe that the active GIH is more highly acetylated than inactive GIHs and possesses more euchromatic physical properties. Taken together these data show that while most GIs were initially euchromatic during the polygenic phase of OR expression, once choice has taken place, GIHs possess distinct OR interaction properties, chromatin marks, and physical features that are determined by their association with the active OR gene. I believe that these data are best explained by a winner-takes-all event, where GIHs containing transcribed OR genes during the polygenic phase are in competition for choice. Once one OR begins to win, it recruits resources to maintain its expression which consequently results in the silencing of other GIHs. Ectopic induction of OR gene transcription is sufficient to bias choice and silence other ORs by impeding their specific association with a GIH. I find that this does not depend on the coding properties of OR protein, as the transcription of non-coding OR mRNAs still results in OR gene silencing. I describe this competition as a symmetry breaking process, where asymmetrical reorganization of transcriptional resources to a single GIH is mediated by non-coding properties of a single highly expressed OR mRNA, culminating in the stable expression of that allele alone for the remainder of a cell’s lifetime.

M.Sc. (Zoology) / Please refer to full text to view abstract

Frequency-dependent selection has long been a popular heuristic explanation for the maintenance of genetic diversity in natural populations. Indeed, a large body of theoretical and empirical work has already gone into elucidating the causes and consequences of frequency-dependent selection. Most theoretical work, to date, has focused either on the diallelic case, or dealt with only very specific forms of frequency-dependence. A general model of the maintenance of multiallelic genetic diversity has been lacking. Here we extend a flexible general model of frequency-dependent selection, the pairwise interaction model, to the case of multiple alleles. First, we investigate the potential for genetic variation under the pairwise interaction model using a parameter-space approach. This approach involves taking a large random sample of all possible fitness sets and initial allele-frequency vectors of the model, iterating each to equilibrium from each set of random initial conditions, and measuring how often variation is maintained, and by which parameter combinations. We find that frequency- dependent selection maintains full polymorphism more often than classic constant-selection models and produces more skewed equilibrium allele frequencies. Fitness sets with some degree of rare advantage maintained full polymorphism most often, but a variety of non-obvious fitness patterns were also found to have positive potential for polymorphism. Second, we further investigate some unusual dynamics uncovered by the parameter-space approach above. Long-period allele-frequency cycles and a small number of aperiodic trajectories were detected. We measured the number, length and domains of attraction of the various attractors produced by the model. The genetic cycles produced by the model did not have periods short enough to be observable on an ecological time scale. In a real world system, allele-frequency cycling is likely to be indistinguishable from stable equilibrium when observed over short time scales. Third, we use a construction approach to model frequency-dependent selection with mutation under the pairwise interaction model. This approach involves the construction of an allelic polymorphism by bombarding an initial monomorphism with mutant alleles over many generations. We find that frequency-dependent selection is able to generate large numbers of alleles at a single locus. The construction process generates a wide range of allele- frequency distributions and genotypic fitness relationships. We find that constructed polymorphisms remain permanently invasible to new mutants. Analysis of constructed fitness sets may even reveal a signature of positive frequency dependence. Finally, we examine the numbers and distributions of fitnesses and alleles produced by construction under the pairwise interaction model with mutation from existing alleles, using several different methods of generating mutant fitnesses. We find that, relative to more general construction models, generating mutants from existing alleles lowers the average number of alleles maintained by frequency-dependent selection. Nevertheless, while the overall numbers of alleles are lower, the polymorphisms produced are more stable, with more natural allele-frequency distributions. Overall, frequency-dependent selection remains a powerful mechanism for the maintenance of genetic variation, although it does not always work in intuitively obvious ways.

Thesis (M.S.)--Rochester Institute of Technology, 2006. / Typescript. Includes bibliographical references (leaves 38-39).

No description available.

The genetic variation among extant life forms reflects the outcomes of evolution. The fodder of evolution – germline mutations – is shaped by the interplay among evolutionary forces – notably natural selection and random genetic drift. In turn, these forces leave footprints recorded in the genetic variation of extant life forms. Characterizing these footprints to understand how evolution works is at the heart of population genetics. To this end, massive datasets of genetic variation have opened new avenues of research, around how mutation rates evolve for instance, and reinvigorated long standing questions in population genetics, notably about the genetic basis of adaptation. In turn, theoretical models of evolution inform what kind of footprints we expect evolution to leave behind in such data. Two theoretical models that investigate open questions in population genetics are described in this thesis. In Chapter 1, I consider the evolution of germline mutation rates, particularly on short evolutionary timescales, and ask if recently observed variation in mutation rates among human lineages could be explained by evolution at genetic modifiers of mutation rates. Genetic modifiers of mutation rates are expected to evolve under purifying selection: mutations at modifiers that increase mutation rates (“mutator alleles”) should be selected against, because they increase the burden of deleterious mutations in individuals who carry them. The frequencies of mutator alleles are also affected by mutation, genetic drift, and demographic processes. We model the evolution of mutator alleles under the interplay of these forces and characterize the dynamics at mutation rate modifiers as a function of the efficacy of selection acting on them. We find that modifiers under intermediate selection have the greatest contribution to variation in mutation rates between distantly related populations, but only variation at strongly selected modifiers turns over fast enough to explain variation in mutation rates among human lineages. We also predict that strongly selected modifiers could be potentially identified in the contemporary datasets of human pedigrees used to study germline mutations. In Chapter 2, I consider a central and enduring question in evolutionary biology: whether adaptation typically arises from few large effect changes or from many small effect changes. Both sides are supported by ample evidence. Yet it is unclear how to translate this evidence into general answers about the genetic basis of adaptation, in part because different methodologies have different limitations and ask different questions. Theory may offer a way out of this quagmire or at least a start. To this end, we reframe the question in terms of traits and ask: how does the genetic basis of adaptation depend on the ecological and genetic attributes of a trait? To start answering this question, I model adaptation in a simple yet highly relevant setting. I consider a trait under stabilizing selection and assume the distribution of trait values in the population is initially at mutation-selection-drift-balance. I then characterize the adaptive response that is elicited by a sudden change in the environment. I find that the adaptive response, and notably the probability that adaptation arises from the fixation of large effect alleles, depends on the size of the environmental change and the genetic architecture of the trait. These attributes are measurable and can be directly related to the disparate evidence that we have about the genetic basis of adaptation. Thus, this kind of modeling may help translate such evidence into general conclusions about how traits evolve. My thesis work was interrupted by the global COVID-19 pandemic, and in response to this pandemic, governments around the world implemented shelter-in-place protocols. However, essential workers were exempt from these protocols, potentially decreasing their efficacy. In Chapter 3, we describe our epidemiological project, aimed at understanding the impact of essential workers on epidemic control. To this end, we model three different archetypes of essential workers under a reasonably realistic SEIR model of the COVID-19 pandemic. We find that the different social interactions that essential workers maintain qualitatively changes their personal risk of infection and the spread of the overall epidemic. These results highlight the utility of not considering essential workers as a monolithic group but instead distinguishing between the impact of different types of essential workers on epidemic control.

Predicting the effect of missense variants is critically important in population and medical genetics. It is essential to interpret genetic variation in population screening and clinical diagnostic sequencing, to reach optimal statistical power of risk gene discovery in genetic studies of diseases and traits. A quantitative analysis of the fitness effect of all possible missense variants can provide a foundation for understanding how proteins evolve in humans and other species. In this thesis, I describe new methods to infer the effect of missense variants using various machine learning techniques. First, I worked on a ResNet-based supervised model to predict pathogenicity trained on curated databases. The curated clinical databases have uneven quality and uncertain bias across genes. To address this issue, I developed a new method, MisFit, to separately model the molecular effect and population fitness effect of missense variants, and to estimate them jointly using a probabilistic graphical model. The architecture of MisFit follows the biological causality of the variant effect, that is, for a missense variant, the protein sequence and structure context determine its molecular effect, which in turn determines its fitness effect given how the protein is involved in various conditions and traits. The latter is a latent factor encapsulated in a sigmoid-shaped function with gene-specific parameters. The fitness effect determines the expected allele counts in human populations. This model can be trained using large-scale population genome data without known pathogenicity labels. I investigated how informative allele counts are for inferring fitness effect using simulations with realistic demographic parameters. To take advantage of the latest deep learning techniques and large population genome data sets, I use a Poisson-Inverse-Gaussian distribution, which is differentiable, to approximate the probability of allele counts given fitness effect and sample size. We show that MisFit estimated heterozygous selection coefficient of missense variants is consistent with ratio of de novo mutations among observed variants in a population with child-parents trio data. Furthermore, de novo missense variants with selection coefficient >0.01 are significantly enriched in neurodevelopmental disorders cases, achieving the best performance in prioritization of de novos for new risk gene discovery compared to previous methods. We also show that the estimated molecular effect reached the state-of-the-art performance in the classification of damaging variants in deep mutational scanning assays, with improved consistency of the score scale across genes. Finally, I analyzed the transmission disequilibrium of inherited variants in autism using a new empirical Bayesian method to identify risk genes, which models relative risk as a continuous function of variant effect in each gene.

Thesis (Ph. D.)--Ohio State University, 2007. / Full text release at OhioLINK's ETD Center delayed at author's request