Perturbation Free-Energy Toolkit: An Automated Alchemical Topology Builder

Abstract

Free-energy calculations play an important role in the application of computational chemistry to a range of fields, including protein biochemistry, rational drug design, or materials science. Importantly, the free-energy difference is directly related to experimentally measurable quantities such as partition and adsorption coefficients, water activity, and binding affinities. Among several techniques aimed at predicting free-energy differences, perturbation approaches, involving the alchemical transformation of one molecule into another through intermediate states, stand out as rigorous methods based on statistical mechanics. However, despite the importance of free-energy calculations, the applicability of the perturbation approaches is still largely impeded by a number of challenges, including the definition of the perturbation path, i.e., alchemical changes leading to the transformation of one molecule to the other. To address this, an automatic perturbation topology builder based on a graph-matching algorithm is developed, which can identify the maximum common substructure (MCS) of two or multiple molecules and provide the perturbation topologies suitable for free-energy calculations using the GROMOS and the GROMACS simulation packages. Various MCS search options are presented leading to alternative definitions of the perturbation pathway. Moreover, perturbation topologies generated using the default multistate MCS search are used to calculate the changes in free energy between lysine and its two post-translational modifications, 3-methyllysine and acetyllysine. The pairwise free-energy calculations performed on this test system led to a cycle closure of 0.5 ± 0.3 and 0.2 ± 0.2 kJ mol –1 , with GROMOS and GROMACS simulation packages, respectively. The same relative free energies between the three states are obtained by employing the enveloping distribution sampling (EDS) approach when compared to the pairwise perturbations. Importantly, this toolkit is made available online as an open-source Python package ( ).

Relative binding free energy calculations with transformato: A molecular dynamics engine-independent tool

“…In the study introducing transformato (Wieder et al, 2022), we demonstrated that one can achieve very high accuracy with the CC/ SAI approach for (relative) solvation free energy differences. Our present results demonstrate that the approach can be extended to large-scale RBFE calculations and that, overall, the agreement with related approaches (Wang et al, 2015;Song et al, 2019;Gapsys et al, 2020;He et al, 2020;Petrov, 2021) is good.…”

“…They require significant computing resources and are comparatively slow; setting up the simulations and analysis procedures is difficult and tedious, even for experts. To make their utilization easier, several front ends to biomolecular simulation packages have been developed to help set up RBFE simulations ( Seeliger and de Groot, 2010 ; Gapsys et al, 2015 ; Loeffler et al, 2015 ; Wang et al, 2015 ; Zhang H. et al, 2021 ; Petrov, 2021 ). Here, we present a related tool, transformato, which, in contrast to most other tools, is not dependent on a particular simulation program.…”

Accelerated Enveloping Distribution Sampling (AEDS) Allows for Efficient Sampling of Orthogonal Degrees of Freedom

“… 50 The different ligands under study were parameterized using the Automated Topology Builder web server (ATB) and adjusted to better match the GROMOS charges. 51 The python package SMArt 52 was used to build the reference state and perturbation topologies using a single topology approach, which contains the necessary extra atoms for all of the endstates as dummy particles. 53 , 54 To simplify the perturbation topology, the bond lengths of the five-member ring of indene, indole, and benzofurane remained constant.…”

Exploring the Effect of Enhanced Sampling on Protein Stability Prediction

“…Several approaches exist for the computational prediction of protein thermostability changes upon single-point mutations, ranging from detailed all-atom molecular dynamics (MD) simulations, simplified energy models, statistical approaches, and phylogenetic analyses to machine learning methods. − MD simulations in explicit solvent evolved toward a promising tool in the prediction of protein thermostability − although the computational expense often still hampers a high-throughput setup. However, this will be more and more facilitated by the increased calculation speed accompanied by pre- and postprocessing tools that make the setup of such calculations more user-friendly ,− as well as the establishment of best practices to set up and analyze such calculations. , The study of Gapsys et al showed that different force fields may perform quite differently in their ability to predict free-energy changes due to side-chain mutations. Based on this finding, they suggested a consensus approach, utilizing predictions in multiple force fields, that, together with increased sampling time, reached a prediction accuracy comparable to those obtained in experiments.…”