Computational alchemy for molecular design and optimization

项目摘要/摘要 药物发现既耗时又昂贵，现在每一种新药都要推向市场。 平均花费远远超过10亿美元。这一成本是由DIFfi对药物发现的狂热推动的，部分原因是 在fi和调节生物分子的功能的初始“HIT”的过程中涉及的试错量， 然后将这些重新fi成对于生物分子靶标和其他所需的具有足够的ffi密度的“先导” 属性。在这里，我们开发和改进计算方法来指导这一过程，允许潜在的 EFfi潜在销售线索在创建之前要经过计算测试-大大减少了 过程中涉及的试错量和指导分子设计过程的量。 在这里，我们建立在小组和fi的先前工作的基础上，根据以下公式计算炼金术自由能 分子模拟--目前最有希望用于指导药物发现的计算技术。然而， 这样的技术只适用于有限的案例子集，需要大量的专业知识才能使用，甚至 他们的局限性还没有被很好地理解。在这里，我们重点扩展系统的范围，可以 用这些技术处理，使计算更健壮和快速，提高准确性，并识别 以及分离剩余的fi序列以供修复。 炼金术自由能计算具有特别的前景，因为它们的准确性和物理上的真实性。 ISM。在这里，我们集中讨论这些计算的技术和应用，重点是(1)提高fi效率 和结合自由能计算的准确性；(2)自由能的自动化和大规模标杆 指导工作以确保稳健性和准确性的计算；以及(3)利用模拟和 自由能计算用于指导潜在抗癌相扑E-1抑制剂的先导发现和优化 毒品。总的来说，目标1-2专注于迭代地改进和测试计算工具，而目标3专注于 与实验合作者一起开发了一个特定的fic应用程序。 这项工作承诺更准确和更快速的自由能计算，范围更广，这样他们就可以 可靠地应用于药物发现和其他领域的分子设计问题。我们的长期工作旨在 制作一个Workflow，化学家开发结合特定靶点的新分子可以输入数百个 一组潜在的化合物，在某一天下班前合成到计算机中，然后返回工作 第二天早上fi和这些化合物根据预测的目标fi密度，选择性， 溶解性和其他性质，使多年的合成和分析得以省去。在这里，我们开发， 测试和应用技术，以帮助使这项工作flow成为可能，建立在我们之前在 用于绑定预测的物理模型。这也利用和扩展了我们之前的R01中构建的技术。

PROJECT SUMMARY/ABSTRACT Pharmaceutical drug discovery is time-consuming and expensive, with each new drug brought to market now costing well over $1 billion on average. This cost is driven by the difficulty of drug discovery, and in part by the amount of trial and error involved in the process of finding initial “hits” which modulate the function of a biomolecule, and then refining these into “leads” which have adequate affinity for the biomolecular target and other desirable properties. Here, we develop and improve computational methods to guide this process, allowing the potential efficacy of prospective leads to be tested computationally prior to their creation — dramatically reducing the amount of trial and error involved in the process and guiding the molecular design process. Here, we build on previous work in the group and the field on alchemical free energy calculations based on molecular simulations — the most promising present computational technique for guiding drug discovery. However, such techniques work well only for a limited subset of cases, require considerable expertise to employ, and even their limitations are not yet well understood. Here, we focus on expanding the range of systems which can be treated with these techniques, making the calculations more robust and rapid, improving accuracy, and identifying and isolating remaining deficiencies for repair. Alchemical free energy calculations hold particular promise both because of their accuracy and physical real- ism. Here, we focus on technology and applications of these calculations, focusing on (1) improved efficiency and accuracy of binding free energy calculations; (2) automation and large-scale benchmarking of free energy calculations to guide work to ensure robustness and accuracy; and (3) applications to utilizing simulations and free energy calculations to guide lead discovery and optimization of SUMO E-1 inhibitors as potential anti-cancer drugs. Broadly, Aims 1-2 focus on iteratively improving and testing computational tools, whereas Aim 3 focuses on a specific application with experimental collaborators. This work promises more accurate and more rapid free energy calculations, with broader scope so that they can reliably be applied to molecular design problems in drug discovery and elsewhere. Our long-term work aims to produce a workflow where a chemist developing new molecules to bind a particular target could input hundreds of potential compounds to synthesize next into a computer before leaving work one day, and return to work the following morning to find these compounds automatically prioritized based on predicted target affinity, selectivity, solubility and other properties, allowing years worth of synthesis and assays to be bypassed. Here, we develop, test and apply technologies to help make this workflow possible, building on our extensive previous success in physical modeling for binding prediction. This also leverages and extends technologies built in our prior R01.