Investigating how non-homologous recombination structures genes, proteins, operons, clusters, genomes and ecosystems

点击翻译按钮获取中文摘要

Since Darwin's time we have thought of the evolution of life on the planet in terms of a great unifying tree of life. Darwin, writing to Thomas Henry Huxley said "The time will come (though I will not live to see it), when we shall have fairly true genealogies of each great kingdom of life". For much of the intervening years, the focus has been on trying to construct this great tree of life. However, while much of life is tree-like and diversifying, much of life is also involved in the process of merging. From simple symbioses, which might or might not become permanent (e.g. the chloroplast that we see powering plant life on the planet is the descendent of a once free-living bacterium), to the hybridization of plants or animals, to the fusion of genes, we see many, many instances of mergings. Unfortunately, our knowledge of these mergings lags well behind our knowledge of diversifying evolution. In this proposal, we will broaden our understanding of mergings and make such analyses easier and more comprehensive.Our first objective is to develop software to help analyse genetic data. In this effort we have been helped enormously - and perhaps quite surprisingly - by entities such as Google, Facebook and Twitter. These companies have based their technology around the kinds of graphs we are using for molecular sequences. When a person joins Facebook, they are represented by the software as a "node" on a graph. When they "friend" somebody, then an "edge" is drawn between these two nodes. When they "like" a post or a page, a different kind of edge is drawn between the person node and the page node. People and pages form a bipartite graph. Pages are characterised, say as being political pages, or pages with an interest in sport or furniture, etc. Therefore, there is another level for pages. Overall, between people, pages, groups, interests, etc. Facebook represents their entire business as a multi-level graph. We are now doing the same kind of thing for evolving entities.In the case of multilevel analysis of evolving objects, we can represent the smallest of evolving objects (say, a protein domain) as a node. If two domains are homologous (they share a common ancestor and are related), then we can draw an edge between them. If the appear on the same protein/gene, then we can draw edges between the domains and that gene (like as if two people have the same interest in fishing, on facebook). We can then characterise the gene as being of a particular "kind", say metabolic, or membrane-embedded. We can also indicate genes on our network, that are sitting on the same chromosome (analogous to saying they have the same "interest"). We can also have a network level where we indicate whether the organism is free-living, pathogenic, anaerobic, involved in a metabolic consortium, etc.In the same way that we see on social networks that communities form, we see on sequence networks that communities form. There are many parallels and we can gain significant insights into how evolution is really structuring life on the planet. For instance, preliminary studies have shown that some sequences are promiscuous and some are not. Certain domains are widespread in genes, while some are only found in one kind of sequence and no other. We see plasmids, such as those found in the Lyme-disease-causing bacterium Borrelia that have unique kinds of genes, but these genes are found across the diversity of Borrelia plasmids. In other words, the genes are species-restricted, but not plasmid restricted.The outcome of this programme will be to have flexible software and several new insights into how evolution has structures genes and genomes.