Statistical Mechanics with Quantum Potentials: Application to Host-Gues

描述（由申请人提供）：宿主分子，如环糊精和葫芦酚，可以“捕获”较小的分子并影响其物理和化学行为。宿主分子附着的越强，即与较小的“客体”结合的程度越高，效果就越大。宿主分子本身也可以被化学改变（即衍生化），这可以改变它们与客体分子结合的强度，以及它们自身的物理性质。科学家们正在为主客技术发现许多与人类健康有关的应用，包括改进药物的特性，使其更有效和更安全，清除化学战剂的潜在清除剂，以及清理环境化学污染物。这一领域的基础研究和应用/工业研发的数量正在迅速增加。给定一个特定的“客体”分子（如候选药物、化学污染物），研发科学家需要的关键信息是宿主-客体结合亲和力和关联/解离率。这个SBIR项目旨在开发一个软件工具，可以准确地预测这些主客结合特性（例如结合自由能）。这将允许研发科学家进行计算实验，减少所需的昂贵和耗时的实验数量。目前需要开发这样一种软件工具，因为正如最近盲法测试挑战所证明的那样，现有工具不够准确，无法为研究人员提供有用的信息。最近的研究表明，将量子力学（QM）能量函数与严格的统计力学相结合，可以显著提高预测的准确性。然而，这些概念验证研究尚未转化为适用于应用研发的强大计算工具。因此，该项目将把VeraChem目前的统计力学软件包（VM2）与广泛使用的量子化学软件包GAMESS结合起来，并实现各种计算方案的驱动程序来实现这一目标。在这种混合方法中，分子构象的玻尔兹曼分布仍将通过彻底的构象搜索产生，就像对经典VM2一样；然而，构象搜索将不仅仅依赖于分子力学，而是由更可靠的QM势来指导。QM势也将用于熵项，包括对非调和效应的处理。将充分利用最近在半经验QM （SEQM）可靠性方面的显着改进，并在更高水平的QM上进行可选的修正。通过并行处理和复杂的共形滤波器/审查过程，计算的周转将加快。

DESCRIPTION (provided by applicant): Host molecules, such as cyclodextrins and cucurbiturils, can 'capture' smaller molecules and affect their physical and chemical behavior. The stronger the host molecule holds onto, i.e. binds, its smaller 'guest' the larger the effect ca be. Host molecules themselves can also be chemically altered (i.e. derivatized), which can change how strongly they bind guest molecules, as well as their own physical properties. Scientists are discovering many human health-related applications for host-guest technology, including improvement of the properties of drugs to make them more effective and safer, potential scavengers for chemical warfare agent removal, and clean-up of environmental chemical pollutants. The amount of basic research as well as applied/industrial R&D in this area is expanding rapidly. Given a particular 'guest' molecule (e.g. drug candidate, chemical pollutant) key pieces of information R&D scientists require is the host-guest binding affinity and the association/dissociation rates. This SBIR project aims to develop a software tool that can accurately predict these host-guest binding properties (e.g. binding free energy). This would allow R&D scientists to carry out computational experiments reducing the number of expensive and time-consuming bench experiments required. There is a current need for such a software tool to be developed, because as recently demonstrated by a blinded test challenge, existing tools are not accurate enough to provide useful information to researchers. Very recent studies indicate that the accuracy of the predictions can be significantly improved by combining quantum mechanical (QM) energy functions with rigorous statistical mechanics. However, these proof-of-concept studies have yet to be translated into a robust computational tool suitable for applied R&D. Therefore, this project will interface VeraChem's current statistical mechanics software package (VM2) with the widely used quantum chemistry package GAMESS, and implement drivers for various computational schemes to achieve this goal. In this proposed hybrid methodology, a Boltzmann distribution of molecular conformations will still be generated via a thorough conformational search as it is for classical VM2; however, the conformational search will not solely rely on molecular mechanics but will be guided by the more reliable QM potential. QM potentials will also be used for entropy terms, including a treatment of anharmonic effects. Full advantage will be taken of recent dramatic improvements in reliability of semi-empirical QM (SEQM), with optional corrections at higher levels of QM. Turnaround of calculations will be speeded up by parallel processing and a sophisticated conformer filter/vetting process.