Characterizing variation in the local structure of fitness landscapes to assess the predictability of evolution

PI：Kryazhimskiy，Sergey 项目总结/摘要 癌症会产生耐药性吗？病毒是否会跨越宿主并引起新的大流行？这些根和 另一个公共卫生问题是进化。支配进化的原则是很好理解的，但我们不能 预测它并回答上面提出的问题。我们可能对两种进化预测感兴趣。 表型如何随时间变化？哪些基因变化会导致这种进化？在许多 种群，适应是由新的突变驱动的。一个主要的挑战是，新的突变对 表型和适应性取决于它们产生的遗传背景。这种依赖关系称为 上位性使表型和遗传水平的进化预测复杂化。 最简单的表型预测是健身。要做到这一点，我们需要知道新的突变如何影响适应性。这 信息包含在称为新突变适应性效应分布（DFE）的量中。上位性 可以导致DFE从一个基因型到另一个基因型的变化，我们不知道它是如何变化的。目标1 本研究旨在预测适应度的演变。为了实现这一目标，我们将测量DFE如何在基因型之间变化， 在进化中产生，并在这种变异中找到平衡。 预测遗传进化更加困难，因为通常有太多不同的适应性突变， 可以在人群中出现。上位性使情况进一步复杂化。如果没有上位性或者上位性是这样的 在一种基因型中有益的所有突变在所有其他基因型中也是有益的， (i.e.,高度可预测的）适应性进化的最终遗传结果，但突变路径的数量 导致它将非常大。预测任何给定的人口选择哪条道路可能是不可能的。 另一方面，如果上位性使某些基因型中有益的突变在其他基因型中有害， 自然选择可利用的突变路径的数量将减少，使其更可预测。 然而，这些路径可能导致不同的最终基因型，使进化的结果更难预测。 布莱。不同类型上位性的流行率尚不清楚。本研究的目标2是了解如何 可预测的突变路径和进化结果。为了实现这一目标，我们将衡量 突变在基因型之间变化，并量化不同类型的上位性。 我们将使用几种新的基因工程方法在实验酵母种群中实现我们的目标， 测序技术。我们将使用条形码谱系跟踪来估计多个酵母菌株中的DFE， 以及与基于条形码测序的适合性测定偶联的化学和插入诱变。我们将测量- 使用一种新的CRISPEY方法，全基因组测序， 适应性突变体和插入诱变的筛选。最后，我们将把这些数据综合到一个“适应度”模型中 景观”并试图预测我们人口的进化。 1

PI: Kryazhimskiy, Sergey Project Summary/Abstract Will a cancer develop drug resistance? Will a virus jump hosts and cause a new pandemic? The root of these and other public health concerns is evolution. The principles that govern evolution are well understood, yet we cannot predict it and answer the questions posed above. We might be interested in two kinds of evolutionary predictions. How would a phenotype change over time? And which genetic changes would cause this evolution? In many populations, adaptation is driven by new mutations. A major challenge is that the effects of new mutations on phenotypes and on fitness depend on the genetic background in which they arise. Such dependencies, called “epistasis”, complicate evolutionary prediction at both the phenotypic and genetic levels. The simplest phenotype to predict is fitness. To do so, we need to know how new mutations affect fitness. This information is contained in the quantity called the distribution of fitness effects of new mutations (DFE). Epistasis can cause the DFE to vary from on genotype to another, and we do not understand how it varies. Objective 1 of this research is predict fitness evolution. To achieve it, we will measure how DFEs vary across genotypes that arise in evolution and find regularities in this variation. Predicting genetic evolution is more difficult because there are usually too many different adaptive mutations that can arise in the population. Epistasis complicates the situation further. If there is no epistasis or if epistasis is such that all mutations beneficial in one genotype are also beneficial in all other genotypes, there would be only one (i.e., highly predictable) eventual genetic outcome of adaptive evolution, but the number of mutational paths leading to it would be very large. Predicting which path any given population takes would likely be impossible. On the other hand, if epistasis makes certain mutations that are beneficial in some genotypes deleterious in others, the number of mutational paths accessible to natural selection would decrease, making them more predictable. However, these paths could lead to different eventual genotypes making the outcomes of evolution less predicta- ble. The prevalence of different types of epistasis is unknown. Objective 2 of this research is to understand how predictable mutational paths and evolutionary outcomes are. To achieve it, we will measure how the effects of mutations vary across genotypes and quantify different types of epistasis. We will approach our objectives in experimental yeast populations using several novel genetic engineering and sequencing techniques. We will estimate the DFEs in multiple yeast strains using barcode lineage tracking, as well as chemical and insertion mutagenesis coupled with barcode sequencing-based fitness assays. We will meas- ure the fitness effects of hundreds of individual mutations using a novel CRISPEY method, full-genome sequenc- ing of adaptive mutants and insertion mutagenesis. Finally, we will synthesize these data in a model of a “fitness landscape” and attempt to predict evolution of our populations. 1