Training machine learning potentials for reactive systems: A Colab tutorial on basic models.

In the last several years, there has been a surge in the development of machine learning potential (MLP) models for describing molecular systems. We are interested in a particular area of this field - the training of system-specific MLPs for reactive systems - with the goal of using these MLPs to accelerate free energy simulations of chemical and enzyme reactions. To help new members in our labs become familiar with the basic techniques, we have put together a self-guided Colab tutorial (https://cc-ats.github.io/mlp_tutorial/), which we expect to be also useful to other young researchers in the community. Our tutorial begins with the introduction of simple feedforward neural network (FNN) and kernel-based (using Gaussian process regression, GPR) models by fitting the two-dimensional Müller-Brown potential. Subsequently, two simple descriptors are presented for extracting features of molecular systems: symmetry functions (including the ANI variant) and embedding neural networks (such as DeepPot-SE). Lastly, these features will be fed into FNN and GPR models to reproduce the energies and forces for the molecular configurations in a Claisen rearrangement reaction.

Computational Evaluation of Potential Molecular Catalysts for Nitrous Oxide Decomposition.

Nitrous oxide (N2O) is a potent greenhouse gas (GHG) with limited use as a mild anesthetic and underdeveloped reactivity. Nitrous oxide splitting (decomposition) is critical to its mitigation as a GHG. Although heterogeneous catalysts for N2O decomposition have been developed, highly efficient, long-lived solid catalysts are still needed, and the details of the catalytic pathways are not well understood. Reported herein is a computational evaluation of three potential molecular (homogeneous) catalysts for N2O splitting, which could aid in the development of more active and robust catalysts and provide deeper mechanistic insights: one Cu(I)-based, [(CF3O)4Al]Cu (A-1), and two Ru(III)-based, Cl(POR)Ru (B-1) and (NTA)Ru (C-1) (POR = porphyrin, NTA = nitrilotriacetate). The structures and energetic viability of potential intermediates and key transition states are evaluated according to a two-stage reaction pathway: (A) deoxygenation (DO), during which a metal-N2O complex undergoes N-O bond cleavage to produce N2 and a metal-oxo species and (B) (di)oxygen evolution (OER), in which the metal-oxo species dimerizes to a dimetal-peroxo complex, followed by conversion to a metal-dioxygen species from which dioxygen dissociates. For the (F-L)Cu(I) activator (A-1), deoxygenation of N2O is facilitated by an O-bound (F-L)Cu-O-N2 or better by a bimetallic N,O-bonded, (F-L)Cu-NNO-Cu(F-L) complex; the resulting copper-oxyl (F-L)Cu-O is converted exergonically to (F-L)Cu-(η2,η2-O2)-Cu(F-L), which leads to dioxygen species (F-L)Cu(η2-O2), that favorably dissociates O2. Key features of the DO/OER process for (POR)ClRu (B-1) include endergonic N2O coordination, facile N2 evolution from LR'u-N2O-RuL to Cl(POR)RuO, moderate barrier coupling of Cl(POR)RuO to peroxo Cl(POR)Ru(O2)Ru(POR)Cl, and eventual O2 dissociation from Cl(POR)Ru(η1-O2), which is nearly thermoneutral. N2O decomposition promoted by (NTA)Ru(III) (C-1) can proceed with exergonic N2O coordination, facile N2 dissociation from (NTA)Ru-ON2 or (NTA)Ru-N2O-Ru(NTA) to form (NTA)Ru-O; dimerization of the (NTA)Ru-oxo species is facile to produce (NTA)Ru-O-O-Ru(NTA), and subsequent OE from the peroxo species is moderately endergonic. Considering the overall energetics, (F-L)Cu and Cl(POR)Ru derivatives are deemed the best candidates for promoting facile N2O decomposition.

Visualization of Electron Density Changes Along Chemical Reaction Pathways.

We propose a simple procedure for visualizing the electron density changes (EDC) during a chemical reaction, which is based on a mapping of rectangular grid points for a stationary structure into (distorted) positions around atoms of another stationary structure. Specifically, during a small step along the minimum energy pathway (MEP), the displacement of each grid point is obtained as a linear combination of the motion of all atoms, with the contribution from each atom scaled by the corresponding Hirshfeld weight. For several reactions (identity SN2, Claisen rearrangement, Diels-Alder reaction, [3+2] cycloaddition, and phenylethyl mercaptan attack on pericosine A), our EDC plots showed an expected reduction of electron densities around severed bonds (or those with the bond-order lowered), with the opposite observed for newly-formed or enhanced chemical bonds. The EDC plots were also shown for copper triflate catalyzed N2O fragmentation, where the N-O bond weakening initially occurred on a singlet surface, but continued on a triplet surface after reaching the minimum-energy crossing point (MECP) between the two potential energy surfaces.

Effect of an Amyloidogenic SARS-COV-2 Protein Fragment on α-Synuclein Monomers and Fibrils.

Aggregates of α-synuclein are thought to be the disease-causing agent in Parkinson's disease. Various case studies have hinted at a correlation between COVID-19 and the onset of Parkinson's disease. For this reason, we use molecular dynamics simulations to study whether amyloidogenic regions in SARS-COV-2 proteins can initiate and modulate aggregation of α-synuclein. As an example, we choose the nine-residue fragment SFYVYSRVK (SK9), located on the C-terminal of the envelope protein of SARS-COV-2. We probe how the presence of SK9 affects the conformational ensemble of α-synuclein monomers and the stability of two resolved fibril polymorphs. We find that the viral protein fragment SK9 may alter α-synuclein amyloid formation by shifting the ensemble toward aggregation-prone and preferentially rod-like fibril seeding conformations. However, SK9 has only a small effect on the stability of pre-existing or newly formed fibrils. A potential mechanism and key residues for potential virus-induced amyloid formation are described.

Effect of an amyloidogenic SARS-COV-2 protein fragment on α-synuclein monomers and fibrils.

Using molecular dynamic simulations we study whether amyloidogenic regions in viral proteins can initiate and modulate formation of α-synuclein aggregates, thought to be the disease-causing agent in Parkinson's Disease. As an example we choose the nine-residue fragment SFYVYSRVK (SK9), located on the C-terminal of the Envelope protein of SARS-COV-2. We probe how the presence of SK9 affects the conformational ensemble of α-synuclein monomers and the stability of two resolved fibril polymorphs. We find that the viral protein fragment SK9 may alter α-synuclein amyloid formation by shifting the ensemble toward aggregation-prone and preferentially rod-like fibril seeding conformations. However, SK9 has only little effect of the stability of pre-existing or newly-formed fibrils.