ATD: DEEP SEQUENCING OF MICROBIAL POPULATIONS: DISENTANGLING DIVERSITY, DYNAMICS, AND ERRORS

微生物种群的深度测序是其多样性及其动态进化历史的潜在有力探针。 不幸的是，分析深度测序数据和在存在错误的情况下推断信息的能力远远赶不上DNA测序能力。在调查密切相关的微生物群体的精细尺度多样性时，困难尤其明显。研究人员和他的同事将开发新的算法，用于提取有关微生物种群基因组多样性的可靠信息，并分析可以产生这种多样性的短期动态进化过程。这些算法将基于对深度测序数据中产生错误和偏差的过程进行建模，主要是DNA的PCR扩增和测序本身。但是，为了对精细尺度的多样性做出有用的推断，需要更好地了解大型微生物种群的进化动力学。 因此，一系列的进化情景将被建模和分析，重点是多样性和它可能给进化历史的线索。然后，为将多样性与错误区分开来而开发的算法将被集中并调整，以使用进化建模的期望作为先验信息，从而区分不同的场景。这将包括为DNA测序的深度、广度和时间制定优化策略。动物的进化通常非常缓慢，但细菌和病毒的进化非常快，这种进化导致对人类的重大威胁。例如，囊性纤维化儿童体内通常无害的细菌的进化最终导致他们过早死亡，而在全球范围内，流感的进化是导致新流行病的原因。迫切需要更好地了解和观察病原体的演变。 在实验室里，细菌和病毒也会进化，而且这种进化是可以被引导的。 虽然人工进化可以带来许多好处，比如细菌可以吃污染物，但它也可以用于邪恶的目的。一个关键的能力，例如对十年前炭疽袭击的调查，是从样本中确定进化历史：它们何时、何地以及如何进化。幸运的是，DNA测序已经变得如此便宜，以至于人们不仅可以对许多单个细菌或病毒进行测序，而且还可以对整个种群进行测序。这使得能够直接观察种群的进化。个体之间的差异谱，提供了自然选择作用的变异，以及进化历史的线索。 但是DNA测序会产生许多错误，这使得提取有用的信息变得非常困难。 该项目将开发新的算法，用于从错误中解开实际的DNA序列。与此同时，复杂的数学模型将用于探索各种可能的进化历史和由此产生的序列变异。这些将被放在一起，以制定最佳使用DNA测序的策略，以推断细菌和病毒种群进化的关键方面，并理解和预测其后果。

Deep sequencing of microbial populations is a potentially powerful probe of their diversity and their dynamical evolutionary history. Unfortunately, the ability to analyze deep sequencing data and to infer information in the presence of errors has far from kept up with DNA sequencing capabilities. The difficulties are particularly pronounced for investigating the fine-scale diversity in a population of closely related microbes. The investigator and his colleagues will develop new algorithms for extraction of reliable information on the genomic diversity of microbial populations and analyze the short-term dynamical evolutionary processes that can generate such diversity. The algorithms will be based on modeling the processes that produce errors and biases in deep sequencing data, primarily the PCR amplification of the DNA and the sequencing itself. But in order to make useful inferences about fine-scale diversity, far better understanding is needed of the evolutionary dynamics of large microbial populations. Thus a spectrum of evolutionary scenarios will be modeled and analyzed focussing on the diversity and clues it may give to the evolutionary history. The algorithms developed for disentangling the diversity from errors will then be focussed and adapted to use the expectations from the evolutionary modeling as prior information and thereby distinguish between different scenarios. This will include developing optimized strategies for depth, breadth, and timing of DNA sequencing.Evolution of animals is usually very slow, but bacteria and viruses evolve extremely fast and this evolution leads to major threats to humans. For example, the evolution of usually innocuous bacteria within children with cystic fibrosis is what eventually leads to their premature death, and, on a global scale, evolution of influenza is what causes new epidemics. Better understanding and observations of evolution of pathogens is sorely needed. In the laboratory, bacteria and viruses also evolve --- and this evolution can be directed. Although artificial evolution can lead to many benefits, such as bacteria that eat pollutants, it can also be used for nefarious purposes. A crucial capability, such as for investigation of the anthrax attacks ten years ago, is to determine the evolutionary history from samples: when, where, and how they evolved. Fortunately, DNA sequencing has become so inexpensive that one can not only sequence many individual bacteria or viruses, but also sequence whole populations. This enables direct observations of the evolution of a population. the spectrum of differences among the individuals that provides the variation on which natural selection acts, and clues to the evolutionary history. But DNA sequencing produces many errors which make extraction of the useful information exceedingly difficult. This project will develop new algorithms for disentangling the actual DNA sequences from the errors. In parallel, sophisticated mathematical modeling will be used to explore various possible evolutionary histories and the resulting sequence variations. These will be put together to develop strategies for optimal use of DNA sequencing for inferring key aspects of the evolution of bacterial and viral populations, and understanding and predicting their consequences.