CORE 1: BIOMEDICAL COMPUTATION RESEARCH

核心1：生物医学计算研究

• 批准号: 8045678
• 负责人: RUSS BIAGIO ALTMAN
• 金额: $ 134.29万
• 依托单位: STANFORD UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2010
• 资助国家: 美国
• 起止时间: 2010-09-30 至 2014-08-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/8045678
• 关键词: Address; Algorithms; Architecture; Area; Arts; Automobile Driving; Binding; Biological; Biomechanics; Biomedical Computing; Biomedical Research; Cardiovascular system; Cell Shape; Cell surface; Cells; Cereals; Classification; Code; Collaborations; Communities; Complex; Computer Hardware; Computer software; Computers; Data; Development; Elements; Engineering; Ensure; Environment; Equation; Evolution; Family; Free Energy; Generations; Goals; Human; Imagery; Language; Laws; Lead; Life; Mechanics; Medical; Methods; Mission; Modeling; Molecular; Molecular Conformation; Motion; Movement; Myosin ATPase; Organism; Output; Pattern; Performance; Physics; Positioning Attribute; Property; Proteins; RNA Folding; Research; Research Personnel; Research Project Grants; Running; Scientist; Simulate; Software Engineering; Source Code; Specialist; Speed; Structure; System; Technology; Testing; Time; Transistors; Translating; Virus Diseases; Work; base; biocomputing; biomedical scientist; design; experience; flexibility; improved; innovation; mathematical model; models and simulation; molecular dynamics; molecular mechanics; neuromuscular; next generation; open source; operation; particle; programs; protein folding; scale up; simulation; success; syntax; tool

CORE 1: BIOMEDICAL COMPUTATION RESEARCH The mission of Simbios is to develop, disseminate, and support a simulation tool kit (SIMTK) that enables biomedical scientists to develop and share accurate models and simulations of biological structures¿from molecules to organisms. We have developed, tested, and released SIMTK version 2.0, which includes high performance algorithms for performing matrix operations, generating and integrating equations of motion, performing linear and nonlinear optimization, modeling contact between bodies, and calculating molecular interaction forces (Figure ES.2). SIMTK has enabled the development of powerful graphics-based applications. For example, OPENSIM, built from SIMTK, is focused on simulation of human biomechanics, using LAPACK for linear algebra, SIMBODY for multibody dynamics, IPOPT for optimization, and other SIMTK components. OPENMM ZEPHYR, an easy-to-use interface for atomistic molecular dynamics, builds on SIMTK using GPU-accelerated molecular force field calculations. RNABUILDER simulates coarse-grained models of large complexes of RNAs and proteins, making extensive use of LAPACK, SIMBODY and other SIMTK components. Through SIMTK, and the applications that use it, we have enabled thousands of researchers to do rigorous, high-performance physics-based simulations. SIMTK has been developed and tested in close collaboration with hundreds of biomedical scientists to ensure its accuracy and utility. Our past driving biological problems (DBPs) have included research projects in RNA folding, protein folding, myosin dynamics, cardiovascular mechanics, and neuromuscular biomechanics. By choosing DBPs that represent important areas of research, our software innovations find broad applications. SIMTK version 2.0 contains two complementary systems: a sophisticated open source multibody mechanics code, SIMBODY, that forms the basis for modeling applications in biomechanics and molecular mechanics, and an interacting particle open source code, OPENMM, that provides extremely fast force-field computations for large numbers of interacting components. These codes are based on state-of-the-art research innovations, and are built and documented by experienced software engineering professionals, who have developed and delivered complex software packages to thousands of users. LEVERAGING A NEW GENERATION OF COMPUTER HARDWARE Physical simulation is one of the most computationally intensive activities in biocomputing, and therefore is highly dependent on advances in hardware technology. Recently, there has been a shift in hardware towards complex heterogeneous multicore architectures. This is not simply computing with graphics cards, but a much more fundamental shift in how Moore's law of computing power will advance: clock rates have stopped improving but the transistors continue to get smaller, and will be arranged in massively parallel arrays on special purpose hardware. We will take the lead in ensuring that biophysical simulation develops appropriately to use these new architectures, and have engaged in a collaboration with the Stanford Pervasive Parallelism Lab (PPL) to design "domain specific languages" (DSLs) that will provide an application programmer interface that hides the complexity of programming these complex new architectures.

核心1：生物医学计算研究 Simbios的使命是开发，传播和支持模拟工具套件（SIMTK），该工具包（SIMTK）使生物医学科学家能够开发和共享从分子到组织的生物结构的准确模型和模拟。我们已经开发了，测试和发布的SIMTK版本2.0，其中包括用于执行矩阵操作，生成和集成运动方程，进行线性和非线性优化的高性能算法，实现物体之间的建模接触以及计算分子相互作用力（图ES.2）。 Simtk已启用了强大的基于图形的应用程序的开发。例如，由Simtk构建的Opensim专注于模拟人类生物力学，使用Lapack 线性代数，用于多体动力学的imbody，用于优化的IPOPT和其他SIMTK组件。 OpenMM Zephyr是一种易于使用的原子分子动力学界面，它使用GPU加速分子力场计算在SIMTK上建立。 RNABUILDER模拟了RNA和蛋白质大型复合物的粗粒模型，从而广泛使用Lapack，Simbody和其他Simtk组件。通过Simtk以及使用它的应用程序，我们使成千上万的研究人员能够进行严格，高性能的物理模拟。 Simtk已与数百位生物医学科学家密切合作开发和测试，以确保其准确性和实用性。我们过去的驾驶生物学问题（DBP）包括RNA折叠，蛋白质折叠，肌球蛋白动力学，心血管力学和神经肌肉生物力学方面的研究项目。通过选择代表重要研究领域的DBP，我们的软件创新找到了广泛的应用。 SIMTK 2.0版包含两个完整的系统：一个复杂的开源多机械力学代码，Simbody，构成了在生物力学和分子力学中建模应用的基础，以及一个相互作用的粒子开源代码，OpenMM，该代码为大量相互作用的组件提供了极其快速的力量计算。这些代码基于最先进的研究创新， 并由经验丰富的软件工程专业人员建造和记录，他们已将复杂的软件包开发给了数千名用户。 利用新一代的计算机硬件 物理仿真是生物计算中计算中最密集的活动之一，因此高度取决于硬件技术的进步。最近，硬件转向复杂的异质多层体系结构。这不仅是使用图形卡进行计算，而且在摩尔的计算法定定律将推动的方式上发生了更大的基本转变：时钟速率已经停止改进，但是晶体管继续变小，并且会在特殊用途硬件上以大量平行的阵列排列。我们将领导着确保生物物理模拟适当地使用这些新体系结构，并与斯坦福大学普遍存在合作 实验室（PPL）设计“特定域”（DSL）的实验室（DSL）将提供一个应用程序程序员界面，该界面隐藏了编程这些复杂的新体系结构的复杂性。