Department of Chemistry, Columbia University, New York, New York 10027, United States.
Department of Systems Biology, Columbia University Medical Center, New York, New York 10032, United States.
J Chem Theory Comput. 2024 Oct 8;20(19):8609-8623. doi: 10.1021/acs.jctc.4c00954. Epub 2024 Sep 27.
Relative binding free energy (RBFE) simulation is a rigorous approach to the calculation of quantitatively accurate binding free energy values for protein-ligand binding in which a reference binder is gradually converted to a target binder through alchemical transformation during the simulation. The success of such simulations relies on being able to accurately sample the correct conformational phase space for each alchemical state; however, this becomes a challenge when a significant conformation change occurs between the reference and target binder-receptor complexes. Increasing the simulation time and using enhanced sampling methods can be helpful, but effects can be limited, especially when the free energy barrier between conformations is high or when the correct target complex conformation is difficult to find and maintain. Current RBFE protocols seed the reference complex structure into every alchemical window of the simulation. In our study, we describe an improved protocol in which the reference structure is seeded into the first half of the alchemical windows, and the target structure is seeded into the second half of the alchemical windows. By applying information about the relevant correct end point conformations to different simulation windows from the beginning, the need for large barrier crossings or simulation prediction of the correct structures during an alchemical simulation is in many cases obviated. In the diverse cases we examine below, the simulations yielded free energy predictions that are satisfactory as compared to experiment and superior to running the simulations utilizing the conventional protocol. The method is straightforward to implement for publicly available FEP workflows.
相对结合自由能 (RBFE) 模拟是一种严格的方法，可用于计算蛋白质-配体结合的定量准确结合自由能值，其中通过模拟过程中的化学转变，逐渐将参考配体转化为目标配体。此类模拟的成功依赖于能够准确地对每个化学态的正确构象相空间进行采样；然而，当参考和目标配体-受体复合物之间发生显著构象变化时，这就成为一个挑战。增加模拟时间和使用增强采样方法可能会有所帮助，但效果可能会受到限制，尤其是当构象之间的自由能势垒较高或难以找到和维持正确的目标复合物构象时。目前的 RBFE 方案将参考复合物结构播种到模拟的每个化学态窗口中。在我们的研究中，我们描述了一种改进的方案，其中将参考结构播种到化学态窗口的前半部分，将目标结构播种到化学态窗口的后半部分。通过从一开始就将有关相关正确终点构象的信息应用于不同的模拟窗口，可以避免在化学模拟过程中需要进行大的势垒穿越或模拟预测正确结构的情况。在我们下面检查的各种情况下，模拟产生的自由能预测与实验相比是令人满意的，并且优于使用传统方案运行模拟的结果。该方法对于公开的 FEP 工作流程来说易于实现。
