Lawrenz Morgan, Baron Riccardo, Wang Yi, McCammon J Andrew
Department of Chemistry and Biochemistry, Center for Theoretical Biological Physics, University of California, San Diego, La Jolla, CA, USA.
Methods Mol Biol. 2012;819:469-86. doi: 10.1007/978-1-61779-465-0_27.
The Independent-Trajectory Thermodynamic Integration (IT-TI) approach for free energy calculation with distributed computing is described. IT-TI utilizes diverse conformational sampling obtained from multiple, independent simulations to obtain more reliable free energy estimates compared to single TI predictions. The latter may significantly under- or over-estimate the binding free energy due to finite sampling. We exemplify the advantages of the IT-TI approach using two distinct cases of protein-ligand binding. In both cases, IT-TI yields distributions of absolute binding free energy estimates that are remarkably centered on the target experimental values. Alternative protocols for the practical and general application of IT-TI calculations are investigated. We highlight a protocol that maximizes predictive power and computational efficiency.
描述了一种用于分布式计算自由能计算的独立轨迹热力学积分（IT-TI）方法。与单一热力学积分预测相比，IT-TI利用从多个独立模拟中获得的不同构象采样来获得更可靠的自由能估计值。由于有限采样，后者可能会显著低估或高估结合自由能。我们通过两个不同的蛋白质-配体结合案例举例说明了IT-TI方法的优势。在这两种情况下，IT-TI产生的绝对结合自由能估计值分布都显著集中在目标实验值上。研究了IT-TI计算实际通用应用的替代方案。我们强调了一种能最大化预测能力和计算效率的方案。
