Understanding biomolecular association from large-scale first principles quantum mechanical simulations

分子间联系是大多数生物学过程的基础，例如健康细胞中细胞分裂过程中DNA和蛋白质之间的复杂相互作用，以及癌症等疾病中的DNA和蛋白质之间的复杂相互作用，或者阿尔茨海默氏症和其他与年龄相关的疾病中淀粉样蛋白的明显不可逆组装。因此，主要研究工作的目标是了解和控制生物分子之间的联系，包括开发预防或治愈病理的药物。其目的是使“健康寿命”与不断增长的寿命一样长，并实现“终身健康和幸福”，这是BBSRC的战略研究重点。然而，我们仍然经常不能准确地预测与实际应用相关的生物分子联系。关键是，生物分子相互作用是由结合时发生的电子重排(例如电荷转移和极化)决定的，这些重排的强度不同，但通常采用的力场方法没有很好地考虑到这一点，参数可根据特定情况进行调整。第一原理的量子力学计算(从头算)克服了这些限制，因为它们显式地包括电子；然而，它们具有随系统大小的陡峭(立方)计算标度，不能用于原子超过几百个的分子。由Skylaris博士和他的合作者开发的ONETEP程序能够克服这一限制，因为它基于一种新的量子理论重新表述，该理论与原子数量成线性关系，而不会损失精度。我们已经用ONETEP对原子多达50,000个的系统进行了计算。该项目将得到勃林格-英格尔海姆(BI)的支持，其目的是通过使用量子计算来处理整个系统，从而克服力场的缺点，而不是以前的尝试，即只有一小部分系统是量子的。自2008年10月以来，BI通过BBSRC ICS计划支持了一名博士生，他向这一目标迈出了第一步，展示了在某些众所周知的蛋白质中，整个系统的量子描述可能是必不可少的，因为与具有力场的相同模拟协议相比，它导致了结合自由能的显着改善。受到这些早期成功的鼓舞，我们希望将这种方法应用于更具挑战性的制药系统，以及正在推出的新工具，因为目前在南安普顿、剑桥和伦敦帝国理工学院工作的3名博士后研究人员增强了ONETEP的功能。这些新工具将包括直接在自洽量子计算中的隐式溶剂化模型，更精确的范德华相互作用，以及直接计算结合自由能的大规模从头算分子动力学模拟技术。这项工作的一个平行部分将采用像Arieh Warshel(南加州大学)、Adrial Mulholland(布里斯托尔)和Jonathan Essex(南安普顿，我们与其合作)等工作者的多热力学循环方法，从而将“经典”和“量子”系统视为不同的热力学状态，并使用严格的自由能变化方法(例如自由能微扰理论)来进行这些状态之间的转换，使用统计验收测试来确保每一次“移动”都是有效的。我们将第一次应用这样的方法，让‘量子’是整个系统，而不是它的一小部分，从而拥有一致的表示。到目前为止，我们非常高兴与BBSRC和BI合作，因为它使我们能够在评估新的和潜在非常强大的模拟技术方面迈出第一步。我们的持续合作将确保对制药业产生直接相关的影响，制药业可能远远超出生物分子关联领域，一直延伸到系统生物学。

Intermolecular association underpins most biological processes, such as the complex interactions between DNA and proteins during cell division in healthy cells but also in diseases such as cancers, or the apparently irreversible assembly of amyloid proteins in Alzheimer's and other age-related diseases. As a consequence, major research effort is targeted towards understanding and controlling biomolecular association, including developing drugs that will prevent or cure pathologies. The aim is to make the 'health span' as long as the ever increasing life span and achieve 'lifelong health and well-being', a strategic research priority of the BBSRC. Nevertheless, we are still often not able to predict biomolecular association with accuracy relevant to real applications. The point is that biomolecular interactions are determined by the electronic rearrangements that take place upon association (e.g. charge transfer and polarisation) which vary in strength, but not taken well into account by the force field approaches that are usually employed, with parameters tuned to particular situations. Quantum mechanical calculations from first principles ('ab initio') overcome these limitations as they include the electrons explicitly; however they have steep (cubic) computational scaling with system size and cannot be used in molecules with more than a few hundred atoms. The ONETEP program, developed by Dr Skylaris and his collaborators, is able to overcome this limitation as it is based on a novel reformulation of quantum theory which scales linearly with the number of atoms without loss of accuracy. We have performed calculations with ONETEP on systems with up to 50,000 atoms. The aim of this project, which will be supported by Boehringer Ingelheim (BI), is to overcome the shortcomings of force fields by using quantum calculations to treat the entire system, unlike previous attempts where only a small part of the system has been quantum. Since October 2008 BI have supported a PhD student through the BBSRC ICS scheme who has made the first steps towards this goal by showing that in certain well-known proteins a quantum description of the whole system can be essential as it leads to significant improvements in free energies of binding when compared to the same simulation protocol with force fields. Encouraged by these early successes, we want to apply this approach on more challenging pharmaceutical systems, together with new tools that are coming out as the functionality of ONETEP is currently enhanced by 3 postdoctoral researchers working in Southampton, Cambridge and Imperial College London. These new tools will include an implicit solvation model directly within the self-consistent quantum calculation, more accurate van der Waals interactions, and techniques for large-scale ab initio molecular dynamics simulations with direct calculation of free energies of binding. A parallel strand of the work will employ the multiple thermodynamic cycles approach of workers such Arieh Warshel (USC), Adrial Mulholland (Bristol) and Jonathan Essex (Southampton, with whom we collaborate) whereby 'classical' and 'quantum' systems are considered as different thermodynamic states and rigorous approaches for free energy changes (e.g. free energy perturbation theory) are used to make the transition between these states, using statistical acceptance tests to ensure that each 'move' is valid. For the first time we will apply such approaches by having the 'quantum' be the entire system, rather than a small portion of it, thus having a consistent representation. It has been a great pleasure so far to work with BBSRC and BI as it has enabled us to make the first steps in evaluating new and potentially very powerful simulation technologies. Our continued collaboration will ensure impact of direct relevance to the pharmaceutical industry which could move well outside the biomolecular association domain all the way to systems biology.