published_or_final_version / abstract / Mathematics / Master / Master of Philosophy

The concept of inclusive fitness plays a key role in much of sociobiology. Yet most theoretical studies concerning the evolution of social behavior circumvent inclusive fitness by mobilizing the concept of frequency dependent individual fitness. Given certain assumptions, it is shown that models based on these two different concepts are dynamically equivalent. The models do differ, however, in bookkeeping methods which are advantageous under different circumstances. A knowledge of these circumstances should prove of value to students of social behavior. It is then shown that evolution acts according to an adaptive landscape based on Hamilton's inclusive fitness in the absence of strong selection and inbreeding. This yields an inclusive fitness analogue to much of traditional population genetics. For example, heterozygote superiority in inclusive fitness yields stable polymorphisms, while intermediate dominance results in fixation of one of the alleles. When individuals do not affect one another's fitnesses, the inclusive fitness topography collapses to one based on individual fitness. A general rule for the evolution of social behavior under intermediate dominance is shown to yield Hamilton's Rule as a special case. Next, a general model for examining the evolution of social behavior is developed which, unlike inclusive fitness models, does not require that benefits received be linear functions of the number of social donors encountered. The subsocial route for the evolution of eusociality in haplodiploid organisms is then examined within the context of this model. Nonlinearities render conditions for frequency independent fixation or loss of sister-helping alleles more stringent than expected from models based on the assumption of linear benefits. In particular, both stable polymorphisms and frequency dependent selective thresholds for sister-helping behavior may commonly obtain.

Typescript. / Thesis (Ph. D.)--University of Hawaii at Manoa, 1976. / Bibliography: leaves 142-145. / Microfiche. / xi, 145 leaves ill

Errata is pasted onto front end-paper. / Bibliography: leaves 166-171. / ix, 171 leaves : ill. ; 30 cm. / Title page, contents and abstract only. The complete thesis in print form is available from the University Library. / This thesis considers the question of fixation probabilities and mean absorption times for alleles when a population is divided into a number of subpopulations with asymmetric migration between the subpopulations. The emphasis of the thesis is on small populations and conservation genetics. Results have important implications for management of remnant subpopulations in order to maintain genetic diversity when migration between the remnant subpopulations is not symmetric. / Thesis (Ph.D.)--University of Adelaide, Dept. of Applied Mathematics, 1999?

published_or_final_version / abstract / Mathematics / Doctoral / Doctor of Philosophy

Typescript. / Thesis (Ph. D.)--University of Hawaii, 1966. / Bibliography: leaves 139-146. / x, 232 l mounted illus., tables (part mounted)

Advances in microarray and sequencing technology enable the era of personalized medicine. With increasing availability of genomic assays, clinicians have started to utilize genetics and gene expression of patients to guide clinical care. Signatures of gene expression and genetic variation in genes have been associated with disease risks and response to clinical treatment. It is therefore not difficult to envision a future where each patient will have clinical care that is optimized based on his or her genetic background and genomic profiles. However, many challenges exist towards the full realization of the potential personalized medicine. The human genome is complex and we have yet to gain a better understanding of how to associate genomic data with phenotype. First, the human genome is very complex: more than 50 million sequence variants and more than 20,000 genes have been reported. Many efforts have been devoted to genome-wide association studies (GWAS) in the last decade, associating common genetic variants with common complex traits and diseases. While many associations have been identified by genome-wide association studies, most of our phenotypic variation remains unexplained, both at the level of the variants involved and the underlying mechanism. Finally, interaction between genetics and environment presents additional layer of complexity governing phenotypic variation. Currently, there is much research developing computational methods to help associate genomic features with phenotypic variation. Modeling techniques such as machine learning have been very useful in uncovering the intricate relationships between genomics and phenotype. Despite some early successes, the performance of most models is disappointing. Many models lack robustness and predictions do not replicate. In addition, many successful models work as a black box, giving good predictions of phenotypic variation but unable to reveal the underlying mechanism. In this thesis I propose two methods addressing this challenge. First, I describe an algorithm that focuses on identifying causal genomic features of phenotype. My approach assumes genomic features predictive of phenotype are more likely to be causal. The algorithm builds models that not only accurately predict the traits, but also uncover molecular mechanisms that are responsible for these traits. . The algorithm gains its power by combining regularized linear regression, causality testing and Bayesian statistics. I demonstrate the application of the algorithm on a yeast dataset, where genotype and gene expression are used to predict drug sensitivity and elucidate the underlying mechanisms. The accuracy and robustness of the algorithm are both evaluated statistically and experimentally validated. The second part of the thesis takes on a much more complicated system: cancer. The availability of genomic and drug sensitivity data of cancer cell lines has recently been made available. The challenge here is not only the increasing complexity of the system (e.g. size of genome), but also the fundamental differences between cancers and tissues. Different cancers or tissues provide different contexts influencing regulatory networks and signaling pathways. In order to account for this, I propose a method to associate contextual genomic features with drug sensitivity. The algorithm is based on information theory, Bayesian statistics, and transfer learning. The algorithm demonstrates the importance of context specificity in predictive modeling of cancer pharmacogenomics. The two complementary algorithms highlight the challenges faced in personalized medicine and the potential solutions. This thesis detailed the results and analysis that demonstrate the importance of causality and context specificity in predictive modeling of drug response, which will be crucial for us towards bringing personalized medicine in practice.

The standard representation of evolutionary relationships is a bifurcating tree. However, many types of genetic exchange, collectively referred to as reticulate evolution, involve processes that cannot be modeled as trees. Increasing genomic data has pointed to the prevalence of reticulate processes, particularly in microorganisms, and underscored the need for new approaches to capture and represent the scale and frequency of these events. This thesis contains results from applying new techniques from applied and computational topology, under the heading topological data analysis, to the problem of characterizing reticulate evolution in molecular sequence data. First, we develop approaches for analyzing sequence data using topology. We propose new topological constructions specific to molecular sequence data that generalize standard constructions such as Vietoris-Rips. We draw on previous work in phylogenetic networks and use homology to provide a quantitative measure of reticulate events. We develop methods for performing statistical inference using topological summary statistics. Next, we apply our approach to several types of molecular sequence data. First, we examine the mosaic genome structure in phages. We recover inconsistencies in existing morphology-based taxonomies, use a network approach to construct a genome-based representation of phage relationships, and identify conserved gene families within phage populations. Second, we study influenza, a common human pathogen. We capture widespread patterns of reassortment, including nonrandom cosegregation of segments and barriers to subtype mixing. In contrast to traditional influenza studies, which focus on the phylogenetic branching patterns of only the two surface-marker proteins, we use whole-genome data to represent influenza molecular relationships. Using this representation, we identify unexpected relationships between divergent influenza subtypes. Finally, we examine a set of pathogenic bacteria. We use two sources of data to measure rates of reticulation in both the core genome and the mobile genome across a range of species. Network approaches are used to represent the population of S. aureus and analyze the spread of antibiotic resistance genes. The presence of antibiotic resistance genes in the human microbiome is investigated.

The aim of this study is to place a wide variety of two-phenotype frequency-dependent selection models into a unified population-genetic framework. This work is used to illuminate the possible genetic constraints that may exist in such models, and to address the question of evolutionary modification of these constraints. The first part of Chapter 1 synthesizes from the literature a general framework for applying a genetic structure to a simple class of two-phenotype models. It shows that genetic constraints may prevent the population from achieving a predicted phenotypic equilibrium, but the population will equilibrate at a point that is as close as possible to the phenotypic equilibrium. The second part of Chapter 1 goes on to ask whether evolutionary modification of the genetic system might be expected to remove these constraints. Chapter 2 provides an example of the application of the framework developed in Chapter 1. It presents re-analysis of a model for the evolution of social behavior by reciprocation (Brown et al. 1982). The genetic results of Chapter 1 apply to this model without modification. I show that Brown et al. were unnecessarily restrictive in their assumptions about the types of genetic systems that support their conclusions. Chapter 3 discusses some models for the evolution of altruism that do not fit the assumptions of Chapter 1, despite their two-phenotype structure. These models violate the fundamental assumption of Chapter 1, this being the way in which individual fitness is derived from the behavioral fitnesses. The first part is a complete, in-depth analysis of diploid sib-sib kin selection. I show that some results from the basic model can be used, provided the behavioral inclusive fitness functions are substituted for the true behavioral fitnesses. The second part is an analysis of the validity of the concept of behavioral structure, as introduced by Michod and Sanderson (1985). I show that this concept is flawed as a general principle. Chapter 4 extends the basic model to the case of sex-allocation evolution. I show how many of the central results of sex-allocation theory can be derived more simply using a two-phenotype framework.

Understanding of a global biological network structure by studying its simple local properties through the well-developed field of graph theory is of interest. In particular, in this research an observed biological network was explored through a simulation study. However, one difficulty in such exploration lies on the fitting of graphical models on biological network data. An Exponential Random Graph Model (ERGM) was considered to determine estimations of the several network attributes of complex biological network data. We also compared the estimates of observed network to our random simulated network for both Markov Chain Monte Carlo Maximum Likelihood Estimation (MCMCMLE) and Maximum Pseudo Likelihood Estimation (MPLE) methods under ERGM. The motivation behind this was to determine how different the observed network could be from a randomly simulated network if the physical numbers of attributes were approximately same. Cut-off points of some common attributes of interest for different order of nodes were determined through simulations. We implemented our method to a known regulatory network database of E. coli. / Department of Mathematical Sciences