Using co-evolution to understand the emergence of bacterial phenotype from proteome variation

点击翻译按钮获取中文摘要

Project Summary/Abstract: A fundamental problem in biology is to understand how the compendium of proteins in an organism (the ‘proteome’) cooperatively interact to create phenotype. Despite considerable experimental and computational advances with respect to defining and inferring protein-protein interactions (PPIs), no method currently exists to infer a hierarchy of protein interactions: that is, how proteins interact to create complexes, pathways, and phenotype. This proposal uses bacteria as a model to develop a novel statistical method that transforms a genome sequence into a hierarchy of protein interaction networks. Key to this approach is the advance that components of variation typically discarded as noise (harboring < 0.01% variance) in fact do contain biologically important information regarding PPIs. Preliminary results illustrate that our statistical method may be an effective multi-scale framework to describe emergent biological function arising from a ‘parts-list’ of proteins. We call our approach Spectral Correlation Analysis of Layered Evolutionary Signals (SCALES); the main thrust of our proposal is testing the experimental validity and robustness of our approach. With respect to validity, we will combine high-throughput molecular genetics with computation to test whether SCALES can accurately infer functions of uncharacterized proteins using P. aeruginosa as a model system. With respect to robustness, we will test whether our results are robust to the genomic feature used for measuring co-variation. Looking to the future of the laboratory, SCALES may be generally useful for understanding hierarchical architectures across different biological systems, spanning proteins to cells to ecosystems. Therefore, we believe this proposal will serve as a critical launching point to explore important concepts central to the focus of the post-genomic era, namely creating novel mathematical frameworks by which to convert the torrent of high- content, complex data being collected into useful and actionable biological knowledge. For defining the vision of the laboratory, natural systems are products of a generative process that is poorly understood—the evolutionary process. Though properly described as random variation and selection, evolution generates remarkably ordered, low-entropy biological systems that execute high-performance functions, are robust to perturbation, and have the capacity to adapt to new functions. It is therefore conceivable that quantitatively understanding design architectures of evolved systems, and how they come to be, may yield a new theoretical foundation of engineering for systems with natural-like properties; namely, the ability to dynamically interact with the environment. The broad vision of the Raman Lab is to elucidate organizational principles that govern the ability of evolved systems to work as well as maintain fitness. We hope to address this problem in a variety of systems subject to component variation and environmental selection. In doing so, our ultimate hope is to create rubrics for designing adaptive systems intelligently.