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.

Bridging semiempirical and ab initio QM/MM potentials by Gaussian process regression and its sparse variants for free energy simulation.

Free energy simulations that employ combined quantum mechanical and molecular mechanical (QM/MM) potentials at ab initio QM (AI) levels are computationally highly demanding. Here, we present a machine-learning-facilitated approach for obtaining AI/MM-quality free energy profiles at the cost of efficient semiempirical QM/MM (SE/MM) methods. Specifically, we use Gaussian process regression (GPR) to learn the potential energy corrections needed for an SE/MM level to match an AI/MM target along the minimum free energy path (MFEP). Force modification using gradients of the GPR potential allows us to improve configurational sampling and update the MFEP. To adaptively train our model, we further employ the sparse variational GP (SVGP) and streaming sparse GPR (SSGPR) methods, which efficiently incorporate previous sample information without significantly increasing the training data size. We applied the QM-(SS)GPR/MM method to the solution-phase SN2 Menshutkin reaction, NH3+CH3Cl→CH3NH3++Cl-, using AM1/MM and B3LYP/6-31+G(d,p)/MM as the base and target levels, respectively. For 4000 configurations sampled along the MFEP, the iteratively optimized AM1-SSGPR-4/MM model reduces the energy error in AM1/MM from 18.2 to 4.4 kcal/mol. Although not explicitly fitting forces, our method also reduces the key internal force errors from 25.5 to 11.1 kcal/mol/Å and from 30.2 to 10.3 kcal/mol/Å for the N-C and C-Cl bonds, respectively. Compared to the uncorrected simulations, the AM1-SSGPR-4/MM method lowers the predicted free energy barrier from 28.7 to 11.7 kcal/mol and decreases the reaction free energy from -12.4 to -41.9 kcal/mol, bringing these results into closer agreement with their AI/MM and experimental benchmarks.

Facilitating ab initio QM/MM free energy simulations by Gaussian process regression with derivative observations.

In combined quantum mechanical and molecular mechanical (QM/MM) free energy simulations, how to synthesize the accuracy of ab initio (AI) methods with the speed of semiempirical (SE) methods for a cost-effective QM treatment remains a long-standing challenge. In this work, we present a machine-learning-facilitated method for obtaining AI/MM-quality free energy profiles through efficient SE/MM simulations. In particular, we use Gaussian process regression (GPR) to learn the energy and force corrections needed for SE/MM to match with AI/MM results during molecular dynamics simulations. Force matching is enabled in our model by including energy derivatives into the observational targets through the extended-kernel formalism. We demonstrate the effectiveness of this method on the solution-phase SN2 Menshutkin reaction using AM1/MM and B3LYP/6-31+G(d,p)/MM as the base and target levels, respectively. Trained on only 80 configurations sampled along the minimum free energy path (MFEP), the resulting GPR model reduces the average energy error in AM1/MM from 18.2 to 5.8 kcal mol-1 for the 4000-sample testing set with the average force error on the QM atoms decreased from 14.6 to 3.7 kcal mol-1 Å-1. Free energy sampling with the GPR corrections applied (AM1-GPR/MM) produces a free energy barrier of 14.4 kcal mol-1 and a reaction free energy of -34.1 kcal mol-1, in closer agreement with the AI/MM benchmarks and experimental results.

Computational Investigation of Substituent Effects on the Fluorescence Wavelengths of Oxyluciferin Analogs.

Oxyluciferin, which is the light emitter for firefly bioluminescence, has been subjected to extensive chemical modifications to tune its emission wavelength and quantum yield. However, the exact mechanisms for various electron-donating and withdrawing groups to perturb the photophysical properties of oxyluciferin analogs are still not fully understood. To elucidate the substituent effects on the fluorescence wavelength of oxyluciferin analogs, we applied the absolutely localized molecular orbitals (ALMO)-based frontier orbital analysis to assess various types of interactions (i.e. permanent electrostatics/exchange repulsion, polarization, occupied-occupied orbital mixing, virtual-virtual orbital mixing, and charge-transfer) between the oxyluciferin and substituent orbitals. We suggested two distinct mechanisms that can lead to red-shifted oxyluciferin emission wavelength, a design objective that can help increase the tissue penetration of bioluminescence emission. Within the first mechanism, an electron-donating group (such as an amino or dimethylamino group) can contribute its highest occupied molecular orbital (HOMO) to an out-of-phase combination with oxyluciferin's HOMO, thus raising the HOMO energy of the substituted analog and narrowing its HOMO-LUMO gap. Alternatively, an electron-withdrawing group (such as a nitro or cyano group) can participate in an in-phase virtual-virtual orbital mixing of fragment LUMOs, thus lowering the LUMO energy of the substituted analog. Such an ALMO-based frontier orbital analysis is expected to lead to intuitive principles for designing analogs of not only the oxyluciferin molecule, but also many other functional dyes.

Oxymatrine promotes hypertrophic scar repair through reduced human scar fibroblast viability, collagen and induced apoptosis via autophagy inhibition.

Scars are common complications of burns and trauma, resulting in mental trauma, physical pain, and a heavy financial burden for patients. Specific and effective anti-scarring drugs are lacking in clinical practice. Phytochemicals are easily accessible, low in toxicity, and have various biological and pharmacological properties. Oxymatrine is a phytochemical that regulates autophagy networks. Autophagy is closely related to the maintenance, activity, differentiation, and life-death of skin fibroblasts during wound repair, which results in pathological scars. We hypothesised that oxymatrine may promote hypertrophic scar repair by inhibiting fibroblast autophagy. In vitro studies showed that inhibition of autophagy by oxymatrine decreased viability and collagen metabolism, and increased apoptosis of human scar fibroblasts (HSFs). In vivo studies showed that inhibition of autophagy by oxymatrine promoted scar repair, resulting in a significantly improved final outcome of the hypertrophic scars, a smaller scar area, decreased epidermal and dermal thickness, and a significant downregulation of CK10, P63, collagen I, α-SMA, and TGF-ß1. In summary, oxymatrine promoted hypertrophic scar repair by decreasing HSF viability and collagen, and inducing apoptosis via autophagy inhibition. This study provides a new perspective on the mechanism of hypertrophic burn scar formation, as well as key scientific data for the application of the phytochemical oxymatrine as a new method for the prevention and treatment of hypertrophic scars.

Affordable Ab Initio Path Integral for Thermodynamic Properties via Molecular Dynamics Simulations Using Semiempirical Reference Potential.

Path integral molecular dynamics (PIMD) is becoming a routinely applied method for incorporating the nuclear quantum effect in computer simulations. However, direct PIMD simulations at an ab initio level of theory are formidably expensive. Using the protonated 1,8-bis(dimethylamino)naphthalene molecule as an example, we show in this work that the computational expense for the intramolecular proton transfer between the two nitrogen atoms can be remarkably reduced by implementing the idea of reference-potential methods. The simulation time can be easily extended to a scale of nanoseconds while maintaining the accuracy on an ab initio level of theory for thermodynamic properties. In addition, postprocessing can be carried out in parallel on massive computer nodes. A 545-fold reduction in the total CPU time can be achieved in this way as compared to a direct PIMD simulation at the same ab initio level of theory.

A simplified charge projection scheme for long-range electrostatics in ab initio QM/MM calculations.

In a previous work [Pan et al., Molecules 23, 2500 (2018)], a charge projection scheme was reported, where outer molecular mechanical (MM) charges [>10 Å from the quantum mechanical (QM) region] were projected onto the electrostatic potential (ESP) grid of the QM region to accurately and efficiently capture long-range electrostatics in ab initio QM/MM calculations. Here, a further simplification to the model is proposed, where the outer MM charges are projected onto inner MM atom positions (instead of ESP grid positions). This enables a representation of the long-range MM electrostatic potential via augmentary charges (AC) on inner MM atoms. Combined with the long-range electrostatic correction function from Cisneros et al. [J. Chem. Phys. 143, 044103 (2015)] to smoothly switch between inner and outer MM regions, this new QM/MM-AC electrostatic model yields accurate and continuous ab initio QM/MM electrostatic energies with a 10 Å cutoff between inner and outer MM regions. This model enables efficient QM/MM cluster calculations with a large number of MM atoms as well as QM/MM calculations with periodic boundary conditions.

Controlling the Cleavage of Carbon-Carbon Bonds To Generate α,α-Difluorobenzyl Carbanions for the Construction of Difluoromethylbenzenes.

Controlling the cleavage of carbon-carbon bonds during a chemical reaction is a substantial challenge; however, synthetic methods that accomplish this objective produce valuable and often unexplored reactivity. We have designed a mild process to generate α,α-difluorobenzyl carbanions in the presence of potassium carbonate by exploiting the cleavage of C-C bonds during the release of trifluoroacetate. The initiating reagent is potassium carbonate, which represents an improvement over existing protocols that require a strong base. Fragmentation studies across substituted arenes and heteroarenes were conducted along with computational analyses to elucidate reactivity trends. Furthermore, the mildly generated α,α-difluorobenzyl carbanions from electron-deficient aromatics and heteroaromatic rings can react with aldehydes to create derivatives of difluoromethylbenzenes, which are valuable synthetic targets.

Accelerated computation of free energy profile at ab initio quantum mechanical/molecular mechanical accuracy via a semi-empirical reference potential. II. Recalibrating semi-empirical parameters with force matching.

An efficient and accurate reference potential simulation protocol is proposed for producing ab initio quantum mechanical/molecular mechanical (AI-QM/MM) quality free energy profiles for chemical reactions in a solvent or macromolecular environment. This protocol involves three stages: (a) using force matching to recalibrate a semi-empirical quantum mechanical (SE-QM) Hamiltonian for the specific reaction under study; (b) employing the recalibrated SE-QM Hamiltonian (in combination with molecular mechanical force fields) as the reference potential to drive umbrella samplings along the reaction pathway; and (c) computing AI-QM/MM energy values for collected configurations from the sampling and performing weighted thermodynamic perturbation to acquire an AI-QM/MM corrected reaction free energy profile. For three model reactions (identity SN2 reaction, Menshutkin reaction, and glycine proton transfer reaction) in aqueous solution and one enzyme reaction (Claisen arrangement in chorismate mutase), our simulations using recalibrated PM3 SE-QM Hamiltonians well reproduced QM/MM free energy profiles at the B3LYP/6-31G* level of theory all within 1 kcal mol-1 with a 20 to 45 fold reduction in the computer time.

Representation of the QM Subsystem for Long-Range Electrostatic Interaction in Non-Periodic Ab Initio QM/MM Calculations.

In QM/MM calculations, it is essential to handle electrostatic interactions between the QM and MM subsystems accurately and efficiently. To achieve maximal efficiency, it is convenient to adopt a hybrid scheme, where the QM electron density is used explicitly in the evaluation of short-range QM/MM electrostatic interactions, while a multipolar representation for the QM electron density is employed to account for the long-range QM/MM electrostatic interactions. In order to avoid energy discontinuity at the cutoffs, which separate the short- and long-range QM/MM electrostatic interactions, a switching function should be utilized to ensure a smooth potential energy surface. In this study, we benchmarked the accuracy of such hybrid embedding schemes for QM/MM electrostatic interactions using different multipolar representations, switching functions and cutoff distances. For test systems (neutral and anionic oxyluciferin in MM (aqueous and enzyme) environments), the best accuracy was acquired with a combination of QM electrostatic potential (ESP) charges and dipoles and two switching functions (long-range electrostatic corrections (LREC) and Switch) in the treatment of long-range QM/MM electrostatics. It allowed us to apply a 10Å distance cutoff and still obtain QM/MM electrostatics/polarization energies within 0.1 kcal/mol and time-dependent density functional theory (TDDFT)/MM vertical excitation energies within 10-3 eV from theoretical reference values.

Free Energy Profile Decomposition Analysis for QM/MM Simulations of Enzymatic Reactions.

In enzyme mechanistic studies and mutant design, it is highly desirable to know the individual residue contributions to the reaction free energy and barrier. In this work, we show that such free energy contributions from each residue can be readily obtained by postprocessing ab initio quantum mechanical molecular mechanical (ai-QM/MM) free energy simulation trajectories. Specifically, through a mean force integration along the minimum free energy pathway, one can obtain the electrostatic, polarization, and van der Waals contributions from each residue to the free energy barrier. Separately, a similar analysis procedure allows us to assess the contribution from different collective variables along the reaction coordinate. The chorismate mutase reaction is used to demonstrate the utilization of these two trajectory analysis tools.

Shikonin promotes hypertrophic scar repair by autophagy of hypertrophic scar-derived fibroblasts.

PURPOSE: To investigate the Shikonin (SHI) induce autophagy of hypertrophic scar-derived fibroblasts (HSFs) and the mechanism of which in repairing hypertrophic scar. METHODS: This study showed that SHI induced autophagy from HSFs and repaired skin scars through the AMPK/mTOR pathway. Alamar Blue and Sirius red were used to identify cell activity and collagen. Electron microscopy, label-free quantitative proteomic analysis, fluorescence and other methods were used to identify autophagy. The differences in the expression of autophagy and AMPK/mTOR pathway-related proteins after SHI treatment were quantitatively analyzed by Western blots. A quantitative real-time polymerase chain reaction assay was used to detect the expression of LC3, AMPK and ULK after adding chloroquine (CQ) autophagy inhibitor. RESULTS: After treatment with SHI for 24 hours, it was found that the viability of HSFs was significantly reduced, the protein expression of LC3-II/LC3-I and Beclin1 increased, while the protein expression of P62 decreased. The expression of phosphorylated AMPK increased and expression of phosphorylated mTOR decreased. After the use of CQ, the cell autophagy caused by SHI was blocked. The key genes LC3 and P62 were then reexamined by immunohistochemistry using a porcine full-thickness burn hypertrophic scar model, and the results verified that SHI could induce autophagy in vivo. CONCLUSIONS: These findings suggested that SHI promoted autophagy of HSFs cells, and the potential mechanism may be related to the AMPK/mTOR signal pathway, which provided new insights for the treatment of hypertrophic scars.

Machine learning based implicit solvent model for aqueous-solution alanine dipeptide molecular dynamics simulations.

Inspired by the recent work from Noé and coworkers on the development of machine learning based implicit solvent model for the simulation of solvated peptides [Chen et al., J. Chem. Phys., 2021, 155, 084101], here we report another investigation of the possibility of using machine learning (ML) techniques to "derive" an implicit solvent model directly from explicit solvent molecular dynamics (MD) simulations. For alanine dipeptide, a machine learning potential (MLP) based on the DeepPot-SE representation of the molecule was trained to capture its interactions with its average solvent environment configuration (ASEC). The predicted forces on the solute deviated only by an RMSD of 0.4 kcal mol-1 Å-1 from the reference values, and the MLP-based free energy surface differed from that obtained from explicit solvent MD simulations by an RMSD of less than 0.9 kcal mol-1. Our MLP training protocol could also accurately reproduce combined quantum mechanical molecular mechanical (QM/MM) forces on the quantum mechanical (QM) solute in ASEC environment, thus enabling the development of accurate ML-based implicit solvent models for ab initio-QM MD simulations. Such ML-based implicit solvent models for QM calculations are cost-effective in both the training stage, where the use of ASEC reduces the number of data points to be labelled, and the inference stage, where the MLP can be evaluated at a relatively small additional cost on top of the QM calculation of the solute.

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.

Perspective: Reference-Potential Methods for the Study of Thermodynamic Properties in Chemical Processes: Theory, Applications, and Pitfalls.

In silico investigations of enzymatic reactions and chemical reactions in condensed phases often suffer from formidable computational costs due to a large number of degrees of freedom and enormous important volume in phase space. Usually, accuracy must be compromised to trade for efficiency by lowering the reliability of the Hamiltonians employed or reducing the sampling time. Reference-potential methods (RPMs) offer an alternative approach to reaching high accuracy of simulation without much loss of efficiency. In this Perspective, we summarize the idea of RPMs and showcase some recent applications. Most importantly, the pitfalls of these methods are also discussed, and remedies to these pitfalls are presented.

Stereoelectronic control in regioselective carbohydrate protection.

Organotin-mediated regioselective protection has been extensively used in organic synthesis for many years. However, the mechanistic origin of the resulting regioselectivity is still not clear. By the comparison of the steric and stereoelectronic effects controlling the geometry of five-membered rings formed from neighboring group participation, from intramolecular acyl group migration, or from orthoester transesterification on pyranoside rings, a theory on the pattern resulting from the reaction with dibutyltin oxide is presented. It is thus suggested that the regioselectivity of organotin-mediated protection is controlled by analogous steric and stereoelectronic effects as in neighboring group participation and acyl group migration, mainly dependent on the stereoelectronic effects of the pyranoside itself, and not related to complex stannylene structures. An organotin protection mechanism is also suggested, emanating from steric and stereoelectronic effects, nucleophilicity, and organotin acyl migration.

Role of p53-dependent placental apoptosis in the reproductive and developmental toxicities of caffeine in rodents.

The aim of the present study was to evaluate the role of placental apoptosis in mediating the reproductive and developmental toxicity of caffeine in rodents. Female Kunming mice were treated with caffeine (60, 120 and 240 mg/kg per day) before and during pregnancy. The conception rate, maternal bodyweight gain, placental weight and indices of fetal developmental, including the rate of intrauterine growth retardation (IUGR; i.e. the actual number of fetuses exhibiting IUGR as a percentage of the total number of fetuses), were determined on gestational day (GD) 18. Female Wistar rats were treated with caffeine (20, 60 and 180 mg/kg per day) from GD11 to GD20. The IUGR rate, maternal plasma angiotensin (Ang) II and prolactin concentrations, placental pathology, expression of angiotensin AT(1) and AT(2) receptors and apoptosis-related proteins were measured on GD20. In mice, caffeine treatment dose-dependently reduced the total conception rate, delayed conception and decreased maternal bodyweight gain, placental weight, fetal bodyweight and fetal body and tail lengths, whereas the IUGR rate was increased. In rats, caffeine treatment dose-dependently decreased placental weight and fetal bodyweight and increased the IUGR rate. Abnormal placental structures and decreased maternal plasma prolactin concentrations were observed following 180 mg/kg per day caffeine treatment, which resulted in increases in renin-angiotensin system (RAS) activity, including maternal plasma AngII concentrations and placental AT(1B) and AT(2) receptor expression, and Bax and p53 expression, but decreases in placental Bcl-2 expression. On the basis of the results of the present study, it appears that caffeine ingestion has detrimental effects on the reproductive system and fetal development in rodents that are associated with chronic activation of the maternal and placental RAS, and induction of p53-dependent placental apoptosis.

Accelerating Ab Initio Quantum Mechanical and Molecular Mechanical (QM/MM) Molecular Dynamics Simulations with Multiple Time Step Integration and a Recalibrated Semiempirical QM/MM Hamiltonian.

Molecular dynamics (MD) simulations employing ab initio quantum mechanical and molecular mechanical (ai-QM/MM) potentials are considered to be the state of the art, but the high computational cost associated with the ai-QM calculations remains a theoretical challenge for their routine application. Here, we present a modified protocol of the multiple time step (MTS) method for accelerating ai-QM/MM MD simulations of condensed-phase reactions. Within a previous MTS protocol [Nam J. Chem. Theory Comput. 2014, 10, 4175], reference forces are evaluated using a low-level (semiempirical QM/MM) Hamiltonian and employed at inner time steps to propagate the nuclear motions. Correction forces, which arise from the force differences between high-level (ai-QM/MM) and low-level Hamiltonians, are applied at outer time steps, where the MTS algorithm allows the time-reversible integration of the correction forces. To increase the outer step size, which is bound by the highest-frequency component in the correction forces, the semiempirical QM Hamiltonian is recalibrated in this work to minimize the magnitude of the correction forces. The remaining high-frequency modes, which are mainly bond stretches involving hydrogen atoms, are then removed from the correction forces. When combined with a Langevin or SIN(R) thermostat, the modified MTS-QM/MM scheme remains robust with an up to 8 (with Langevin) or 10 fs (with SIN(R)) outer time step (with 1 fs inner time steps) for the chorismate mutase system. This leads to an over 5-fold speedup over standard ai-QM/MM simulations, without sacrificing the accuracy in the predicted free energy profile of the reaction.

Elucidation of the methyl transfer mechanism catalyzed by chalcone O-methyltransferase: a density functional study.

The mechanism of the methyl transfer catalyzed by chalcone O-methyltransferase has been computationally investigated by employing the hybrid functional B3LYP. Two models are constructed based on the two conformations of the substrate isoliquiritigenin in the X-ray structure. Our calculations show that the overall reaction is divided into two elementary steps: the water-assisted deprotonation of the substrate by His278 as a catalytic base, followed by the methyl transfer from S-adenosyl-L-methionine (SAM) to the substrate. The calculated rate-limiting barriers for the methyl-transfer step indicate that the catalytic reactions are energetically feasible for both conformations adopted by the substrate.

Li1.2Mn0.54Ni0.13Co0.13O2 nanosheets with porous structure as a high-performance cathode material for lithium-ion batteries.

The morphological and structural optimizations of electrode materials are efficient ways to enhance their electrochemical performance. Herein, we report a facile co-precipitation and subsequent calcination method to fabricate Li1.2Mn0.54Ni0.13Co0.13O2 nanosheets consisting of interconnected primary nanoparticles and open holes through the full thickness. By comparing the nanosheets and the agglomerated nanoparticles, the effects of the morphology and structure on the electrochemical performance are investigated. Specifically, the nanosheets exhibit a discharge capacity of 210 mA h g-1 at 0.5C with a capacity retention of 85% after 100 cycles. The improved electrochemical performance could be attributed to their morphological and structural improvements, which may facilitate sufficient electrolyte contacts, short diffusion paths and good structural integrity during the charge/discharge process. This work provides a feasible approach to fabricate lithium-rich layered oxide cathode materials with 2D morphology and porous structure, and reveals the relationships between their morphology, structure and electrochemical performance.