Competition, diversity and success: understanding ecology with genomic data

细菌的基因组数据现在被大量地、连续地收集起来。虽然大型基因组研究本质上主要是描述性的，但这些数据首次提供了在高分辨率下建立和测试进化和生态学定量模型的机会。这是“发现基金”研究项目的长期愿景：我计划将基因组研究从仅仅描述的领域转移出来，开发一个基因组数据可以帮助建立生态现象定量描述的研究体系，并用于完善和检验假设。在接下来的五年里，我将专注于一个关键的不确定领域，即细菌如何构建它们的种群并创造和保持多样性。我的目标是弥合模型和基因组数据之间的差距，将系统基因组数据与建模和统计联系起来，以了解三种不同细菌模型系统中多样性的创造和维持。未来五年有三个中心目标：(1)在一系列尺度上确定种群结构，并直接从系统基因组数据中定义有意义的亚群；(2)描述群体间和群体内的竞争并量化多样性，建立与基因组数据相匹配的可解释模型；(3)利用这些模型进行短期预测，从而用连续采样数据检验模型。我将使用来自三个细菌系统的数据，包括系统发育数据（沿着系统发育树的下降进化）和“非树状”数据（不包括水平转移的基因、聚合位点、表型数据和元数据）。这些在这里被用作自然系统的模型，在可观察的时间尺度上，在存在选择的情况下进化。可以获得这些系统的数据，部分原因是它们与人类健康有关，但这里的重点是基本现象。这种方法有四个关键优势：(1)全基因组序列（WGS）数据的使用将使我们能够在不同的分辨率水平上工作，从小的细粒度尺度到更广泛的人群；(2)我们可以检测和建模惯性模式和自适应模式，因为我们有选择的信息；(3)我们在这三个不同的系统中都有或多或少的“树状”数据，因此我们可以用非树状数据补充系统发育方法；(4)随着时间的推移，数据正在收集中，这为随着时间的推移建立生态过程模型提供了机会，使模型与系统基因组数据相匹配，并测试模型预测。

Genomic data for bacteria are now being gathered in large quantities and serially through time. While large genomic studies have been primarily descriptive in nature, these data present the opportunity to build and test quantitative models in evolution and ecology, at a high level of resolution, for the first time. This is the long-term vision of this Discovery Grant research program: I plan to move genomic studies out of the description-only realm, developing a body of research in which genomic data can help to build quantitative descriptions of ecological phenomena and are used to refine and test hypotheses. In the next five years, I will focus on a key area of uncertainty, namely how bacteria structure their populations and create and maintain diversity. I aim to bridge the gap between models and genomic data, connecting phylogenomic data with modelling and statistics to understand the creation and maintenance of diversity in three distinct bacterial model systems. There are three central objectives over the next five years: (1) to identify population structure at a range of scales and define meaningful subgroups directly from phylogenomic data; (2) to characterize inter- and intra-group competition and quantify diversity, building interpretable models that match genomic data, and (3) to use the models to make short-term predictions and thereby test models with serially sampled data. I will use data from three bacterial systems, comprising both phylogenetic data (that evolves by descent along a phylogenetic tree) and "non tree-like" data (data that do not, including genes that are transferred horizontally, convergent loci, phenotypic data and metadata). These are used here as models of natural systems that are evolving on observable time scales in the presence of selection. Data for these systems are available in part due to relevance for human health, but the focus here is on fundamental phenomena. There are four key advantages to this approach: (1) the use of whole-genome sequence (WGS) data will enable us to work at different levels of resolution, from small fine-grained scales to broader populations; (2) we can detect and model both inertial and adaptive patterns, as we have information about selection; (3) we have both more and less "tree-like" data in these three distinct systems so we can complement phylogenetic methods with non-tree-like data; and (4) data are now being gathered over time, affording the opportunity to model ecological processes through time, fit models to phylogenomic data and test model predictions.