Algorithmic developments to unravel the evolution of viruses and other microbes

点击翻译按钮获取中文摘要

Our research program focuses on developing methods to address outstanding questions in ecology and evolution. As such, our contributions range from microbes to seabirds, with a particular emphasis on viral evolution. Seemingly disparate, these contributions are all united by the design and implementation of ingenious algorithms to address novel biological questions. This proposal will further develop this theme by addressing 4 main hypotheses/research questions. First, we recently developed an algorithm to detect correlated evolution (epistasis) from genetic data. We plan to further develop this tool to analyze multiple genes (and perform whole-genome scans), assess how knowledge of protein structure can guide our algorithm, allow the analysis of multi-state characters (4 for DNA, 20 for proteins) and multiway interactions. With a collaborator, we will test some of the predictions in an in vitro system (Pseudomonas aeruginosa). Second, we recently uncovered a link between selection and epistasis in the Ebola virus. An immediate question is: how general is this? For this, we will mine large datasets to contrast viruses in which selection and epistasis are unlikely (double stranded DNA viruses) with viruses where both selection and epistasis are expected (unsegmented RNA viruses). Third, our recent progress in employing artificial intelligence (machine learning) to address biological questions is motivating us to explore this area further. We will explore ways to better understand evolutionary rate change in influenza viruses, and employ two bacterial models to identify the genomic determinants of phenotypic traits (host restrictions, drug resistance, mercury detoxification) and even fitness. Fourth, we will develop an approach to reconstruct viral transmission networks, employing Approximate Bayes Computing, a technique developed in population genetics but never used in this context. Simulations and a large data set from the Swiss HIV epidemics will be used to validate and test our predictive power. The end-goal of this direction is to provide us with a tool that can assess changes in public health policies. Altogether, this highly interdisciplinary program will provide us with tools that will not only increase our understanding of fundamental evolutionary processes (epistasis & selection), but also help us manage public health policies and help with bioremediation at sites contaminated with mercury. Knowledge will be transferred by means of publications, conference presentations and seminars. End users will range from academics to policy makers (Environment, Public Health). Over the next 5 years, this program will also train at least eight HQP (four undergraduates + four graduate students) in state-of-the-art methods in computational biology, using large datasets ("Big Data", but with a careful design) and state-of-the-art computing environments (local cluster, HPCVL).