Integrating innovative computational and organic synthesis for efficient asymmetric catalyst discovery

背景和目标。开发新的合成方法或分子（例如，新的/更好的催化剂）通常需要长期，迭代和成本效益低的发现和开发过程。与核磁共振光谱、质谱和其他常规使用的技术相比，计算化学很少是有机化学工具箱的一部分。在这种情况下，我们的研究计划侧重于有机合成和计算化学的整合，以提高催化剂和药物发现的成功率。为此，我们开发并应用了我们的平台VIRTUAL CHEMIST（用于设计不对称催化剂）和FORECASTER（用于发现生物活性分子）。由于所有的软件开发和应用程序都是在我们的实验室中完成的，因此反馈回路可以不断改进软件和分子的开发。我们建议进一步开发这些工具，并将其应用于新型不对称催化剂和共价酶抑制剂的设计和合成。计算化学。虚拟化学家和预报员依赖于分子力学（MM）的速度，反过来，使用QM导出的参数，其准确性。随着计算能力的不断提高，一些计算现在可以在QM级别完成，而机器学习（ML）技术可以用来进一步提高准确性。因此，我们建议纳入更多的QM功能以提高准确性（预测立体化学结果，催化活性和药物反应性）以及ML技术，通过ML优化目前使用的力场来改善MM预测，并指导筛选分子的选择。 有机/药物化学。这些工具的应用程序建立的反应（如，石环氧化）已经开始，并将进一步证明计算预测的准确性和它们在不对称催化剂的设计中的使用。我们建议将新的QM和ML功能应用于其他反应，包括具有未知机制和/或没有不对称版本的新型反应。对于这些，我们将1）使用有机和计算方法的组合研究反应机理，2）设计/发现导致改进的立体选择性的催化剂，3）合成和测试这些催化剂。类似地，虽然我们的共价对接程序已被用于开发非常有效的共价抑制剂，但它们的反应性将使用计算预测、合成和测试的组合来改进。所有这些湿实验室实验不仅会产生新的分子，而且会为软件开发人员提供进一步改进的信息。冲击该研究计划将说明集成计算/实验化学的好处，以及有机化学家将计算工具集成到工具箱中的范式转变。HQP将接受高要求技术的培训，如QM和ML以及高级有机合成。

Background and objectives. Developing novel synthetic methodologies or molecules (eg, new/better catalysts) often require long, iterative, and cost-ineffective discovery and development processes. In contrast to NMR spectroscopy, mass spectrometry and other techniques used routinely, computational chemistry is rarely part of the organic chemistry toolbox. In this context, our research program focuses on the integration of synthetic organic and computational chemistry to increase the success rate in both catalyst and drug discovery. To do so, we develop and apply our platforms VIRTUAL CHEMIST (for the design of asymmetric catalysts) and FORECASTER (for the discovery of bioactive molecules). As all the software development and applications are done in our labs, the feedback loop enables the development of constantly improved software and molecules. We propose to further develop these tools and apply them to the design and synthesis of novel asymmetric catalysts and covalent enzyme inhibitors. Computational chemistry. VIRTUAL CHEMIST and FORECASTER rely on molecular mechanics (MM) for its speed which, in turn, uses QM-derived parameters, for their accuracy. With the ever-increasing computational power, some of the computations could now be done at the QM level while machine learning (ML) techniques could be exploited to further improve accuracy. Thus, we propose to incorporate more QM functionalities for improved accuracy (prediction of stereochemical outcome, catalytic activity and reactivity of drugs) as well as ML techniques to improve the MM predictions through ML optimization of the force fields currently used and guide the selection of molecules for screening. Organic/medicinal chemistry. Applications of these tools to established reactions (eg, Shi epoxidation) have started and will further demonstrate the accuracy of the computational predictions and their use in the design of asymmetric catalysts. We propose to apply the new QM and ML features to additional reactions including novel reactions with unknown mechanisms and/or no asymmetric versions. For these, we will 1) investigate the reaction mechanisms using a combination of organic and computational methods, 2) design/discover catalysts leading to improved stereoselectivity and 3) synthesize and test these catalysts. Similarly, while our covalent docking program has been used to develop very potent covalent inhibitors, their reactivity will be refined using a combination of computational prediction, synthesis and testing. All of these wet-lab experiments will not only produce novel molecules but provide information to the software developers for further improvements. Impact. This research program will illustrate the benefit of integrated computational/experimental chemistry and the paradigm shift for organic chemists to integrate computational tools in their toolbox. HQP will be trained on highly demanded techniques such as QM and ML as well as advanced organic synthesis.