Quantum mechanical methods for in silico drug design: Force predictions via machine learning with applications to molecular dynamics simulations

用于计算机药物设计的量子力学方法：通过机器学习进行力预测并应用于分子动力学模拟

• 批准号: 1946486
• 金额: --
• 依托单位: University of Manchester
• 依托单位国家: 英国
• 项目类别: Studentship
• 财政年份: 2017
• 资助国家: 英国
• 起止时间: 2017 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=studentship-1946486
• 关键词: Quantum; mechanical; methods; silico; drug

This project aims to develop high accuracy computational tools for the prediction of ligand conformations, combining quantum mechanics and machine learning methods. The ability to accurately predict ligand conformation is a key limitation in computer-aided drug design approaches. We will leverage machine learning methods for their ability to capture, at a fraction of the computational cost, the high level quantum chemical calculation of subtle physics governing diverse chemical shape. These methods will be applied using state-of-the-art neural networks and supercomputing, with validation against experimental data sets of druglike molecules.

该项目旨在结合量子力学和机器学习方法，开发用于预测配体构象的高精度计算工具。准确预测配体构象的能力是计算机辅助药物设计方法中的关键限制。我们将利用机器学习方法，以计算成本的一小部分来捕获控制不同化学形状的微妙物理的高级量子化学计算。这些方法将使用最先进的神经网络和超级计算来应用，并根据类药物分子的实验数据集进行验证。