Collaborative Research: Development and Application of the AMOEBA Polarizable Force Field

圣路易斯华盛顿大学的Jay Ponder和圣路易斯大学的David Gohara获得了化学系化学理论、模型和计算方法项目以及网络基础设施办公室颁发的奖项，以提高阿米巴(原子多极优化能量生物分子应用)力场模型的分子模拟精度，并通过底层Tinker分子建模软件的并行化来增加模型的适用范围。在这个项目中，研究人员为阿米巴-2模型开发了新的术语，该模型包含了原始阿米巴力场中缺失的物理效应。特别是，电荷转移、穿透效应和衰减的短程色散项被参数化，以消除剩余的最大系统误差。将得到的短程能量模型与量子力学计算进行比较，可以正确地将内部能量划分为不同的项，并为改进阿米巴-2提供一致的基础。现有的修补程序代码以及新的能源术语正在使用混合的OpenMP和MPI范例进行并行化。由此产生的新的潜力和软件将被应用于两个当前需要广泛模拟和高精度水平的问题：有机晶体结构的预测和主客体络合物中配体结合的计算。分子力场计算是计算生物物理学、生物大分子结构确定、分子设计、分子和材料性质模拟等领域的重要工具。在该项目下开发的方法、算法和软件正在通过定期发布TINKER分子建模包的新版本而提供给更广泛的社区。该组件广泛应用于学术研究实验室以及世界各地的化工和制药行业。该小组还通过大学课程参与教育活动，并为希望在自己的教学中使用TINKER软件的教师提供教程材料和实验室练习。

Jay Ponder of Washington University in Saint Louis and David Gohara of Saint Louis University are supported by an award from the Chemical Theory, Models and Computational Methods program in the Division of Chemistry, and the Office of Cyberinfrastructure to improve the accuracy of the AMOEBA (Atomic Multipole Optimized Energetics for Biomolecular Applications) force field model for molecular simulation and to increase the range of applicability of the model via parallelization of the underlying TINKER molecular modeling software. In this project, the researchers develop new terms for an AMOEBA-2 model that encompass physical effects missing from the original AMOEBA force field. In particular, charge transfer, penetration effects and damped short-range dispersion terms are being parameterized to remove the largest remaining systematic inaccuracies. Comparison of the resulting short-range energetic model against quantum mechanical calculations allow correct partitioning of internal energy into the various terms, and provide a consistent basis for improving AMOEBA-2. The existing TINKER code as well as the new energy terms are being parallelized using a hybrid OpenMP and MPI paradigm. The resulting new potential and software will be applied to two current problems that require both extensive simulation and a high level of accuracy: prediction of organic crystal structures, and computation of ligand binding in host-guest complexes. Both of these problem areas have long represented "holy grail" applications for the field of computational chemistry.Molecular force field calculations are important tools in computational biophysics, structure determination of bio-macromolecules, molecular design, and modeling of the properties of molecules and materials. Methods, algorithms and software developed under this project are being made available to the broader community via periodic release of new versions of the TINKER molecular modeling package. This package is widely used in academic research labs and in the chemical and pharmaceutical industries worldwide. The team is also involved in educational activities through University courses, as well as the provision of tutorial materials and laboratory exercises for instructors wishing to use the TINKER software in their own teaching.