A general framework for modelling functional divergence at the molecular level, and investigating relationships to phenotype.

点击翻译按钮获取中文摘要

Proteins participate in an extraordinarily diverse set of chemical reactions and physical tasks, and they have an essential role to play in virtually all biological processes. For this reason there is interest within many research disciplines to understand the origin of protein functions, and to learn how a change in protein-coding DNA relates to differences in biological processes and structures. The latter question is sometimes referred to as "the genotype-phenotype relationship", and understanding this relationship requires knowledge of complex genetic and phenotypic interactions. Recent suggestions that as much as 80% of human non-coding DNA (previously thought to represent "junk" DNA) may also be related to function, only adds to the complexity of the problem. The central question is to what extent can we understand and eventually predict a given biological phenotype from our knowledge of genes and molecules. Through modelling of DNA sequence change, researchers have begun to explore the complexities of many genetic systems. Modelling protein-coding DNA has proved to be a powerful way to investigate how proteins change over time, and how this change might be related to function. However, existing models for protein-coding DNA have limitations, and to make better progress towards our central goal we must continue to improve them. Even more importantly, we need to develop new frameworks for assessing functional divergence that can be generalized across different types of DNA data (not just protein-coding) and different modes of functional divergence (episodic "> This research program will develop several new families of models for detecting functional divergence from DNA sequence data. The new models are united by the common goal of detecting sites within DNA where evolution proceeds via changes (or "switches") in the intensity of natural selection over time. This phenomenon is a hallmark of change in functional constraint, and I propose to model it as a Markov-modulated Markovian process. With the new models in place I can then build an innovative framework for jointly modelling both phenotypic change and DNA sequence evolution. These models will allow us to test for subsets of DNA sites whose evolution is correlated with changes in phenotype. To start, we will focus on a phenotype associated with a single protein. However, through a long-term program of model development, and application to real data, I envision the analyses of higher-level phenotypes such as the pattern of gene expression, or the evolution of pathogen phenotypes (research fields of strategic importance to Canada). I also plan to extend it to the metagenomic setting.