23 February 2023 in Journal of Chemical Theory and Computation
This work presents a variant of an electrostatic embedding scheme that allows the embedding of arbitrary machine learned potentials trained on molecular systems in vacuo. The scheme is based on physically motivated models of electronic density and polarizability, resulting in a generic model without relying on an exhaustive training set. The scheme only requires in vacuo single point QM calculations to provide training densities and molecular dipolar polarizabilities. As an example, the scheme is applied to create an embedding model for the QM7 data set using Gaussian Process Regression with only 445 reference atomic environments. The model was tested on the SARS-CoV-2 protease complex with PF-00835231, resulting in a predicted embedding energy RMSE of 2 kcal/mol, compared to explicit DFT/MM calculations.
28 May 2024 in Journal of Chemical Theory and Computation
We present in this work the emle-engine package (https://github.com/chemle/emle-engine)─the implementation of a new machine learning embedding scheme for hybrid machine learning potential/molecular-mechanics (ML/MM) dynamics simulations. The package is based on an embedding scheme that uses a physics-based model of the electronic density and induction with a handful of tunable parameters derived from in vacuo properties of the subsystem to be embedded. This scheme is completely independent of the in vacuo potential and requires only the positions of the atoms of the machine learning subsystem and the positions and partial charges of the molecular mechanics environment. These characteristics allow emle-engine to be employed in existing QM/MM software. We demonstrate that the implemented electrostatic machine learning embedding scheme (named EMLE) is stable in enhanced sampling molecular dynamics simulations. Through the calculation of free energy surfaces of alanine dipeptide in water with two different ML options for the in vacuo potential and three embedding models, we test the performance of EMLE. When compared to the reference DFT/MM surface, the EMLE embedding is clearly superior to the MM one based on fixed partial charges. The configurational dependence of the electronic density and the inclusion of the induction energy introduced by the EMLE model leads to a systematic reduction in the average error of the free energy surface when compared to MM embedding. By enabling the usage of EMLE embedding in practical ML/MM simulations, emle-engine will make it possible to accurately model systems and processes that feature significant variations in the charge distribution of the ML subsystem and/or the interacting environment.
27 March 2023 in Journal of the American Chemical Society
4 September 2017 in WIREs Computational Molecular Science
19 January 2025 in WIREs Computational Molecular Science
The study of natural enzyme catalytic processes at a molecular level can provide essential information for a rational design of new enzymes, to be applied in more efficient and environmentally friendly industrial processes. The use of computational tools, combined with experimental techniques, is providing outstanding milestones in the last decades. However, apart from the complexity associated with the nature of these large and flexible biomolecular machines, the full enzyme catalyzed process involves different physical and chemical steps. Consequently, from the computational point of view, a deep understanding of every single step requires the selection of a proper computational technique to get reliable, robust and useful results. In this article, we summarize the different computational techniques and their use in the study of every single step of the catalytic process, including conformational diversity, allostery and those to study the chemical steps, as well as in the design of new enzymes. Because of the impact of artificial intelligence in all aspects of science during the last years, special attention has been applied to methods based on these techniques, their foundations and some selected recent applications.
30 September 2024 in Journal of Chemical Information and Modeling
Plastic-degrading enzymes, particularly poly(ethylene terephthalate) (PET) hydrolases, have garnered significant attention in recent years as potential eco-friendly solutions for recycling plastic waste. However, understanding of their PET-degrading activity and influencing factors remains incomplete, impeding the development of uniform approaches for enhancing PET hydrolases for industrial applications. A key aspect of PET hydrolase engineering is optimizing the PET-hydrolysis reaction by lowering the associated free energy barrier. However, inconsistent findings have complicated these efforts. Therefore, our goal is to elucidate various aspects of enzymatic PET degradation by means of quantum mechanics/molecular mechanics (QM/MM) reaction simulations and analysis, focusing on the initial reaction step, acylation, in two thermophilic PET hydrolases, LCC and PES-H1, along with their highly active variants, LCCIG and PES-H1FY. Our findings highlight the impact of semiempirical QM methods on proton transfer energies, affecting the distinction between a two-step reaction involving a metastable tetrahedral intermediate and a one-step reaction. Moreover, we uncovered a concerted conformational change involving the orientation of the PET benzene ring, altering its interaction with the side-chain of the "wobbling" tryptophan from T-stacking to parallel π-π interactions, a phenomenon overlooked in prior research. Our study thus enhances the understanding of the acylation mechanism of PET hydrolases, in particular by characterizing it for the first time for the promising PES-H1FY using QM/MM simulations. It also provides insights into selecting a suitable QM method and a reaction coordinate, valuable for future studies on PET degradation processes.