NEURON2GPU: Transformative Neural Simulation Using Desktop GPU Technology

NEURON2GPU：使用桌面 GPU 技术的变革性神经模拟

• 批准号: 8217134
• 负责人: COSTA M COLBERT
• 金额: $ 34.73万
• 依托单位: INFORMED SIMPLIFICATIONS, LLC
• 依托单位国家: 美国
• 财政年份: 2011
• 资助国家: 美国
• 起止时间: 2011-01-31 至 2013-12-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/8217134
• 关键词: Algorithms; Back; Basic Science; Behavioral Research; Biochemistry; Biological Neural Networks; Biophysics; Brain; Brain Diseases; Businesses; Code; Complement; Computer Hardware; Computer software; Computers; Data; Data Analyses; Development; Devices; Disease; Education; Electronics; Electrophysiology (science); Engineering; Environment; Feasibility Studies; Funding; Future; Genetic Programming; Goals; Health; Human; Image; Language; Libraries; Licensing; Manufacturer Name; Marketing; Mathematics; Measures; Membrane; Modeling; Neurons; Neurosciences; Performance; Pharmacology; Phase; Physiology; Price; Program Development; Province; Research; Research Personnel; Resources; Sales; Scientist; Screening procedure; Small Business Innovation Research Grant; Specialist; Speed; Students; Technology; Testing; Time; Vendor; Work; base; brain research; commercialization; computational neuroscience; cost; drug discovery; experience; feeding; improved; interest; prototype; public health relevance; relating to nervous system; research study; response; scientific computing; simulation; simulation software; software development; supercomputer; technology development; tool

DESCRIPTION (provided by applicant): Researchers are now able to perform computational physiology and pharmacology experiments on realistic model neurons with accurate geometries and full complements of membrane biophysics and biochemistry; however, the practical utility of such simulations are restricted because of the limitations of current commodity computer hardware. We will provide real time, or better, simulations of realistic single neurons and clusters of neurons for drug discovery and physiology research by implementing neural simulation software on low-cost GPU-based desktop supercomputers. For the first time, our product will put the ability to quickly generate parameter landscapes of realistic neural and neural network function on the desks of students and scientists participating in basic research and drug discovery. Our technology will enable physiologists to use online modeling during imaging and electrophysiology experiments via predicting and tuning appropriate experimental manipulations. In its fully developed form, the mature technology will provide realistic neural simulations - across parameter space - which then will be used to modify the parametric control of an ongoing experiment. PUBLIC HEALTH RELEVANCE: We propose to change the landscape of computational neuroscience by putting high-throughput simulation software on the desktops of researchers at low cost. This work will enable not just advances in basic neuroscience leading to a better understanding of how the brain works in health and disease, but it will also provide tools for principled drug discovery for the treatment of human brain disease.

描述(申请人提供)：研究人员现在能够在真实的模型神经元上进行计算生理学和药理学实验，具有准确的几何形状和膜生物物理和生物化学的完整补充；然而，由于当前商品计算机硬件的限制，此类模拟的实际用途受到限制。我们将通过在基于低成本GPU的台式超级计算机上实现神经模拟软件，为药物开发和生理学研究提供实时或更好的真实单个神经元和神经元簇的模拟。我们的产品将首次将快速生成真实神经和神经网络功能的参数景观的能力放在参与基础研究和药物发现的学生和科学家的办公桌上。我们的技术将使生理学家能够通过预测和调整适当的实验操作，在成像和电生理学实验期间使用在线建模。在其完全开发的形式下，这项成熟的技术将提供真实的神经模拟-跨参数空间-然后将被用于修改正在进行的实验的参数控制。 与公共健康相关：我们建议通过以低成本将高通量模拟软件安装在研究人员的桌面上来改变计算神经科学的格局。这项工作不仅将使基础神经科学的进步导致更好地了解大脑在健康和疾病中的工作方式，而且还将为治疗人类大脑疾病的原则性药物发现提供工具。