Combined quantum mechanics/molecular mechanics (QM/MM) Monte Carlo free energy simulations: a feasibility study

尽管科学在进步，每年仍有数百万人死于不治之症。不幸的是，药物开发的成本如此之高，以至于医学研究的重点是有利可图的西方疾病。为了降低开发新药的成本，我们希望能够使用计算机来模拟潜在药物在体内的作用，并利用这些知识来设计新的更好的药物。构建这样的计算模型是具有挑战性的，需要在将足够的细节放入模型以获得现实行为之间保持微妙的平衡，并使模型尽可能简单，以便运行计算不会花费太长时间。到目前为止，使用的大多数模型都非常简单，将药物的原子建模为弹簧上的球。通过将原子视为实心球，模型忽略了原子在化学上最重要的部分，即电子。这是一个严重的疏忽，因为正是电子的相互作用决定了药物是否可以溶解在你的血液中，进入你的细胞，并结合到任何攻击细菌或病毒的蛋白质上，从而中和它们。用量子力学模拟分子中的电子是可能的。然而，使用量子力学对整个蛋白质/药物系统进行建模在计算上过于昂贵。我们建议研究使用量子力学来模拟药物分子的一部分和附近的电子。蛋白质的其余部分仍然可以通过简单的球和弹簧模型来处理，以使计算成为可能。我们将开发的新方法增加了重要的额外细节，使它们更加真实，能够更好地模拟药物如何相互作用。与此同时，这种综合方法应该意味着计算是切实可行的。使我们计划的工作不同的是，它将涉及专门为药物设计定制的混合模型的开发。为这种用途创建一个混合模型将需要克服重大挑战，并开发新的方法来处理模型的量子力学部分与弹簧球部分之间的相互作用。

Despite the advances of science, millions of people still die every year from incurable diseases. Unfortunately, the costs of drug development are so high that the focus of medicinal research is into profitable Western diseases. To reduce the costs of developing new medicinal drugs, we would like to be able to use computers to model how a potential drug works within the body, and to use this knowledge to design new and better drugs. Building computational models like this is challenging, requiring a delicate balance between putting enough detail into the model to get realistic behaviour, and making the model as simple as possible so that it doesn't take too long to run the calculations. Until now, the majority of models used have been very simple, modelling the atoms of a drug as balls on springs. By treating the atoms as solid balls, the models neglect the atom's most chemically important part, namely the electrons. This is a severe oversight, as it is the interactions of electrons that determine whether the drug could dissolve in your blood, work its way into your cells, and bind to, and thus neutralize, the proteins of any attacking bacteria or virus. It is possible to model electrons in molecules using quantum mechanics. However, to model the entire protein/drug system using quantum mechanics would be too computationally expensive. We propose to research the use of quantum mechanics to model just the electrons that are part of, and near to, the drug molecule. The rest of the protein can still be treated by simple ball and springs models to make the calculations possible. The new methods we will develop add important extra detail, making them more realistic and better able to model how drugs interact. At the same time, this combined approach should mean that the calculations are practical to do. What makes our planned work different is that it will involve the development of a mixed model specifically tailored for medicinal drug design. Creating a mixed model for this use will require that significant challenges are overcome, and that new ways are developed to handle the interactions between the quantum mechanics part of the model with the ball on springs part.