ABI Innovation: Domain Architecture Simulator

仿真是现代科学和工程的基本工具。它使物理上不可能的实验成为可能，允许在严格控制的条件下进行测试，并且是确保科学过程中再现性的基本技术。在蛋白质进化中，模拟是理解在超过人类寿命的时间尺度上作用的过程的关键。二十年的研究已经产生了丰富的算法和软件来模拟蛋白质序列进化。然而，目前的模型没有捕捉到多结构域蛋白质进化的过程。多结构域蛋白质是编码蛋白质模块的序列片段的镶嵌体：结构或功能单位称为结构域。由于其模块化的性质，多结构域蛋白质在与其他细胞和环境的细胞通讯中起着核心作用。在人类健康中，多结构域家族与组织修复、细胞凋亡、炎症反应和先天免疫有关。由于大约30%的细菌蛋白质和50%的脊椎动物蛋白质包含两个或两个以上的结构域，模拟的关键好处目前还不能利用至少三分之一的蛋白质在这个星球上。本项目通过设计和实现模拟多结构域进化的软件来解决这一差距。该项目通过开发和分发第一个多结构域蛋白质进化模拟器来推进研究基础设施。该项目还有助于通过本科生和课堂教学的研究经验，建立一个广泛包容的科学工作队伍。本科研究人员将从卡内基梅隆大学本科课程的女性中招募，并通过波多黎各的教育活动从STEM中代表性不足的群体中招募。模拟是分子进化中必不可少的技术，其中感兴趣的过程通常作用于人类奋进无法达到的时间尺度上。仿真对于进化算法和软件的比较和验证、评估竞争假设以及评估系统对扰动的响应是必不可少的。尽管增长的复杂的方法来模拟氨基酸序列的进化，没有模拟器目前存在的结构域插入，复制和删除的过程中，多结构域蛋白质家族的演变模型。虽然这些过程有可能产生任何域的组合，但在自然界中只观察到一小部分可能的域组合，这表明域顺序和共现受到严格限制。这个项目的目标是开发基于这些约束的现实模型的多域仿真算法和原型软件。该模拟器受益于马尔可夫链蒙特卡罗技术和数据驱动的事件概率估计的新组合。数据驱动的事件模块可以快速轻松地重新部署模拟器，以用于不同的分类和蛋白质功能背景。经验测试和模拟器重新设计的重复循环将促进更好地理解自然界的蛋白质设计原理，更精确的模型和方法，以及更准确的进化推论。由此产生的软件将在www.example.com上提供http://www.cs.cmu.edu/_durand/Lab/multidomain.html.This奖项反映了NSF的法定使命，并被认为值得通过使用基金会的知识价值和更广泛的影响审查标准进行评估来支持。

Simulation is a fundamental tool in modern science and engineering. It enables experiments that would otherwise be physically impossible, allows for testing under rigorously controlled conditions, and is an essential technique for ensuring reproducibility in the scientific process. In protein evolution, simulation is the key to understanding processes that act on time scales that exceed a human life span. Twenty years of research have produced a rich repertoire of algorithms and software for simulating protein sequence evolution. However, current models do not capture the processes by which multidomain proteins evolve. Multidomain proteins are mosaics of sequence fragments that encode protein modules: structural or functional units called domains. Because of their modular nature, multidomain proteins play central roles in cellular communication with other cells and with the environment. In human health, multidomain families are implicated in tissue repair, apoptosis, inflammation response, and innate immunity. Since roughly 30% of bacterial proteins and 50% of vertebrate proteins contain two or more domains, the critical benefits of simulation currently cannot be exploited for at least a third of all proteins on this planet.This project addresses this gap through the design and implementation of software for simulating multidomain evolution. This project advances research infrastructure through the development and distribution of the first evolutionary simulator for multidomain proteins. This project also contributes to building a broadly inclusive scientific work force through research experiences for undergraduates and classroom teaching. Undergraduate researchers will be recruited from women in Carnegie Mellon's undergraduate program and from groups that are underrepresented in STEM through educational activities at the University of Puerto Rico. Simulation is an essential technique in molecular evolution, where the processes of interest typically act on the time scales that are beyond the reach of human endeavor. Simulation is essential for comparison and validation of evolutionary algorithms and software, for evaluating competing hypotheses, and for assessing how a system will respond to perturbations. Despite the growth of sophisticated methodology for simulating the evolution of amino acid sequences, no simulators currently exist that model the process of domain insertion, duplication, and deletion by which multidomain protein families evolve. Although these processes have the potential to generate any combination of domains, only a tiny fraction of possible domain combinations are observed in nature, suggesting that domain order and co-occurrence are stringently constrained. The goal of this project is the development of algorithms and prototype software for multidomain simulation based on realistic models of these constraints. This simulator benefits from a novel combination of Markov chain Monte Carlo technology and data-driven estimates of event probabilities. The data-driven event module enables quick and easy redeployment of the simulator for use in different taxonomic and protein function contexts. Repeated cycles of empirical testing and simulator redesign will promote better understanding of nature's protein design principles, more precise models and methods, and more accurate evolutionary inferences. The resulting software will be available at http://www.cs.cmu.edu/_durand/Lab/multidomain.html.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.