Genetic Determinants Of Phenotypic Evolution In Saccharomyces Yeast PDF Download

Phenotypic variation among individuals within populations is ubiquitous in the natural world, and a preeminent challenge in biology is understanding the contribution of genetic variation to this phenotypic variation. Despite technological advances in the development of genome-scale methods for querying molecular phenotypes, our understanding of the molecular basis of morphological and physiological variation remains rudimentary. In this dissertation, I outline computational methods I have developed and analyses I have conducted in the yeast Saccharomyces cerevisiae to make inferences about the relationship between DNA sequences and the molecular phenotypes to which they give rise. First, I describe a population genomics study of a class of genomic elements, intron splice sequences, in a diverse set of complete S. cerevisiae genomes. I obtained quantitative estimates of the strength of purifying selection acting on these sequences, and present analyses suggesting that introns in some subsets of genes are actively maintained in natural populations of S. cerevisiae. Next, I shift my focus to the genetic basis of variation in a particular molecular phenotype, gene expression. I examine genes that show allele-specific expression (ASE) due to cis-regulatory variation, and present a Bayesian statistical model for quantifying ASE measured by RNA-Seq. A novel feature of this model is the ability to detect variable ASE, where the level of ASE differs across a transcript, as can occur in the case of variations in transcript structure. Finally, I explore molecular phenotypic variation more comprehensively, presenting the results of an analysis of deeply phenotyped S. cerevisiae strains. I analyze genome sequence, gene expression, protein abundance, metabolite abundance, and cellular morphological phenotypes in this phenomics study. I identify abundant natural variation across all phenotypic classes, pinpoint loci that act in cis to affect RNA and protein levels, and provide initial clues as to the predictability of phenotypic traits that vary between individuals within a species. I conclude by discussing the need for new statistical models to make use of the rich information contained in functional genomics datasets and the necessity of considering environmental context when disentangling the functional consequences of genetic variation.