Research Center for Analytical Sciences, College of Chemistry, Tianjin Key Laboratory of Biosensing and Molecular Recognition , Nankai University , Tianjin 300071 , China.
Collaborative Innovation Center of Chemical Science and Engineering (Tianjin) , Tianjin 300071 , China.
J Chem Inf Model. 2018 Jul 23;58(7):1315-1318. doi: 10.1021/acs.jcim.8b00115. Epub 2018 Jun 18.
Extended adaptive biasing force (eABF), a collective variable (CV)-based importance-sampling algorithm, has proven to be very robust and efficient compared with the original ABF algorithm. Its implementation in Colvars, a software addition to molecular dynamics (MD) engines, is, however, currently limited to NAMD and LAMMPS. To broaden the scope of eABF and its variants, like its generalized form (egABF), and make them available to other MD engines, e.g., GROMACS, AMBER, CP2K, and openMM, we present a PLUMED-based implementation, called extended-Lagrangian free energy calculation (ELF). This implementation can be used as a stand-alone gradient estimator for other CV-based sampling algorithms, such as temperature-accelerated MD (TAMD) and extended-Lagrangian metadynamics (MtD). ELF provides the end user with a convenient framework to help select the best-suited importance-sampling algorithm for a given application without any commitment to a particular MD engine.
扩展的自适应偏置力（eABF）是一种基于整体变量（CV）的重要性抽样算法，与原始 ABF 算法相比，它被证明非常强大和高效。然而，Colvars 中对其的实现（分子动力学（MD）引擎的软件附加组件）目前仅限于 NAMD 和 LAMMPS。为了拓宽 eABF 及其变体（如广义形式（egABF））的范围，并使其可用于其他 MD 引擎，例如 GROMACS、AMBER、CP2K 和 openMM，我们提出了一种基于 PLUMED 的实现，称为扩展拉格朗日自由能计算（ELF）。此实现可用作其他基于 CV 的采样算法（如温度加速 MD（TAMD）和扩展拉格朗日元动力学（MtD））的独立梯度估计器。ELF 为最终用户提供了一个方便的框架，可帮助选择最适合给定应用的重要性抽样算法，而无需承诺使用特定的 MD 引擎。
