Enhancing molecular design efficiency: Uniting language models and generative networks with genetic algorithms.

This study examines the effectiveness of generative models in drug discovery, material science, and polymer science, aiming to overcome constraints associated with traditional inverse design methods relying on heuristic rules. Generative models generate synthetic data resembling real data, enabling deep learning model training without extensive labeled datasets. They prove valuable in creating virtual libraries of molecules for material science and facilitating drug discovery by generating molecules with specific properties. While generative adversarial networks (GANs) are explored for these purposes, mode collapse restricts their efficacy, limiting novel structure variability. To address this, we introduce a masked language model (LM) inspired by natural language processing. Although LMs alone can have inherent limitations, we propose a hybrid architecture combining LMs and GANs to efficiently generate new molecules, demonstrating superior performance over standalone masked LMs, particularly for smaller population sizes. This hybrid LM-GAN architecture enhances efficiency in optimizing properties and generating novel samples.

Variations in proton transfer pathways and energetics on pristine and defect-rich quartz surfaces in water: Insights into the bimodal acidities of quartz.

HYPOTHESIS: Understanding the mechanisms of proton transfer on quartz surfaces in water is critical for a range of processes in geochemical, environmental, and materials sciences. The wide range of surface acidities (>9 pKa units) found on the ubiquitous mineral quartz is caused by the structural variations of surface silanol groups. Molecular scale simulations provide essential tools for elucidating the origin of site-specific surface acidities. SIMULATIONS: We used density-functional tight-binding-based molecular dynamics combined with rare-event metadynamics simulations to probe the mechanisms of deprotonation reactions from ten representative surface silanol groups found on both pristine and defect-rich quartz (101) surfaces with Si vacancies. FINDINGS: The results show that deprotonation is a highly dynamic process where both the surface hydroxyls and bridging oxygen atoms serve as the proton acceptors, in addition to water. Deprotonation of embedded silanols through intrasurface proton transfer exhibited lower pKa values with less H-bond participation and higher energy barriers, suggesting a new mechanism to explain the bimodal acidity observed on quartz surface. Defect sites, recently shown to comprise a significant portion of the quartz (101) surface, diversify the coordination and local H-bonding environments of the surface silanols, changing both the deprotonation pathways and energetics, leading to a wider range of pKa values (2.4 to 11.5) than that observed on pristine quartz surface (10.4 and 12.1).

Deep learning workflow for the inverse design of molecules with specific optoelectronic properties.

The inverse design of novel molecules with a desirable optoelectronic property requires consideration of the vast chemical spaces associated with varying chemical composition and molecular size. First principles-based property predictions have become increasingly helpful for assisting the selection of promising candidate chemical species for subsequent experimental validation. However, a brute-force computational screening of the entire chemical space is decidedly impossible. To alleviate the computational burden and accelerate rational molecular design, we here present an iterative deep learning workflow that combines (i) the density-functional tight-binding method for dynamic generation of property training data, (ii) a graph convolutional neural network surrogate model for rapid and reliable predictions of chemical and physical properties, and (iii) a masked language model. As proof of principle, we employ our workflow in the iterative generation of novel molecules with a target energy gap between the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO).

Multipole Expansion of Atomic Electron Density Fluctuation Interactions in the Density-Functional Tight-Binding Method.

The accuracy of the density-functional tight-binding (DFTB) method in describing noncovalent interactions is limited due to its reliance on monopole-based spherical charge densities. In this study, we present a multipole-extended second-order DFTB (mDFTB2) method that takes into account atomic dipole and quadrupole interactions. Furthermore, we combine the multipole expansion with the monopole-based third-order contribution, resulting in the mDFTB3 method. To assess the accuracy of mDFTB2 and mDFTB3, we evaluate their performance in describing noncovalent interactions, proton transfer barriers, and dipole moments. Our benchmark results show promising improvements even when using the existing electronic parameters optimized for the original DFTB3 model. Both mDFTB2 and mDFTB3 outperform their monopole-based counterparts, DFTB2 and DFTB3, in terms of accuracy. While mDFTB2 and mDFTB3 perform comparably for neutral and positively charged systems, mDFTB3 exhibits superior performance over mDFTB2 when dealing with negatively charged systems and proton transfers. Overall, the incorporation of the multipole expansion significantly enhances the accuracy of the DFTB method in describing noncovalent interactions and proton transfers.

Quantum biological insights into CRISPR-Cas9 sgRNA efficiency from explainable-AI driven feature engineering.

CRISPR-Cas9 tools have transformed genetic manipulation capabilities in the laboratory. Empirical rules-of-thumb have been developed for only a narrow range of model organisms, and mechanistic underpinnings for sgRNA efficiency remain poorly understood. This work establishes a novel feature set and new public resource, produced with quantum chemical tensors, for interpreting and predicting sgRNA efficiency. Feature engineering for sgRNA efficiency is performed using an explainable-artificial intelligence model: iterative Random Forest (iRF). By encoding quantitative attributes of position-specific sequences for Escherichia coli sgRNAs, we identify important traits for sgRNA design in bacterial species. Additionally, we show that expanding positional encoding to quantum descriptors of base-pair, dimer, trimer, and tetramer sequences captures intricate interactions in local and neighboring nucleotides of the target DNA. These features highlight variation in CRISPR-Cas9 sgRNA dynamics between E. coli and H. sapiens genomes. These novel encodings of sgRNAs enhance our understanding of the elaborate quantum biological processes involved in CRISPR-Cas9 machinery.

Two excited-state datasets for quantum chemical UV-vis spectra of organic molecules.

We present two open-source datasets that provide time-dependent density-functional tight-binding (TD-DFTB) electronic excitation spectra of organic molecules. These datasets represent predictions of UV-vis absorption spectra performed on optimized geometries of the molecules in their electronic ground state. The GDB-9-Ex dataset contains a subset of 96,766 organic molecules from the original open-source GDB-9 dataset. The ORNL_AISD-Ex dataset consists of 10,502,904 organic molecules that contain between 5 and 71 non-hydrogen atoms. The data reveals the close correlation between the magnitude of the gaps between the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO), and the excitation energy of the lowest singlet excited state energies quantitatively. The chemical variability of the large number of molecules was examined with a topological fingerprint estimation based on extended-connectivity fingerprints (ECFPs) followed by uniform manifold approximation and projection (UMAP) for dimension reduction. Both datasets were generated using the DFTB+ software on the "Andes" cluster of the Oak Ridge Leadership Computing Facility (OLCF).

Toward Quantum Chemical Free Energy Simulations of Platinum Nanoparticles on Titania Support.

Platinum nanoparticles (Pt-NPs) supported on titania surfaces are costly but indispensable heterogeneous catalysts because of their highly effective and selective catalytic properties. Therefore, it is vital to understand their physicochemical processes during catalysis to optimize their use and to further develop better catalysts. However, simulating these dynamic processes is challenging due to the need for a reliable quantum chemical method to describe chemical bond breaking and bond formation during the processes but, at the same time, fast enough to sample a large number of configurations required to compute the corresponding free energy surfaces. Density functional theory (DFT) is often used to explore Pt-NPs; nonetheless, it is usually limited to some minimum-energy reaction pathways on static potential energy surfaces because of its high computational cost. We report here a combination of the density functional tight binding (DFTB) method as a fast but reliable approximation to DFT, the steered molecular dynamics (SMD) technique, and the Jarzynski equality to construct free energy surfaces of the temperature-dependent diffusion and growth of platinum particles on a titania surface. In particular, we present the parametrization for Pt-X (X = Pt, Ti, or O) interactions in the framework of the second-order DFTB method, using a previous parametrization for titania as a basis. The optimized parameter set was used to simulate the surface diffusion of a single platinum atom (Pt1) and the growth of Pt6 from Pt5 and Pt1 on the rutile (110) surface at three different temperatures (T = 400, 600, 800 K). The free energy profile was constructed by using over a hundred SMD trajectories for each process. We found that increasing the temperature has a minimal effect on the formation free energy; nevertheless, it significantly reduces the free energy barrier of Pt atom migration on the TiO2 surface and the transition state (TS) of its deposition. In a concluding remark, the methodology opens the pathway to quantum chemical free energy simulations of Pt-NPs' temperature-dependent growth and other transformation processes on the titania support.

Adaptive language model training for molecular design.

The vast size of chemical space necessitates computational approaches to automate and accelerate the design of molecular sequences to guide experimental efforts for drug discovery. Genetic algorithms provide a useful framework to incrementally generate molecules by applying mutations to known chemical structures. Recently, masked language models have been applied to automate the mutation process by leveraging large compound libraries to learn commonly occurring chemical sequences (i.e., using tokenization) and predict rearrangements (i.e., using mask prediction). Here, we consider how language models can be adapted to improve molecule generation for different optimization tasks. We use two different generation strategies for comparison, fixed and adaptive. The fixed strategy uses a pre-trained model to generate mutations; the adaptive strategy trains the language model on each new generation of molecules selected for target properties during optimization. Our results show that the adaptive strategy allows the language model to more closely fit the distribution of molecules in the population. Therefore, for enhanced fitness optimization, we suggest the use of the fixed strategy during an initial phase followed by the use of the adaptive strategy. We demonstrate the impact of adaptive training by searching for molecules that optimize both heuristic metrics, drug-likeness and synthesizability, as well as predicted protein binding affinity from a surrogate model. Our results show that the adaptive strategy provides a significant improvement in fitness optimization compared to the fixed pre-trained model, empowering the application of language models to molecular design tasks.

Effect of surface functional groups on MXene conductivity.

We report the in-plane electron transport in the MXenes (i.e., within the MXene layers) as a function of composition using the density-functional tight-binding method, in conjunction with the non-equilibrium Green's functions technique. Our study reveals that all MXene compositions have a linear relationship between current and voltage at lower potentials, indicating their metallic character. However, the magnitude of the current at a given voltage (conductivity) has different trends among different compositions. For example, MXenes without any surface terminations (Ti3C2) exhibit higher conductivity compared to MXenes with surface functionalization. Among the MXenes with -O and -OH termination, those with -O surface termination have lower conductivity than the ones with -OH surface terminations. Interestingly, conductivity changes with the ratio of -O and -OH on the MXene surface. Our calculated I-V curves and their conductivities correlate well with transmission functions and the electronic density of states around the Fermi level. The surface composition-dependent conductivity of the MXenes provides a path to tune the in-plane conductivity for enhanced pseudocapacitive performance.

Accelerating the density-functional tight-binding method using graphical processing units.

Acceleration of the density-functional tight-binding (DFTB) method on single and multiple graphical processing units (GPUs) was accomplished using the MAGMA linear algebra library. Two major computational bottlenecks of DFTB ground-state calculations were addressed in our implementation: the Hamiltonian matrix diagonalization and the density matrix construction. The code was implemented and benchmarked on two different computer systems: (1) the SUMMIT IBM Power9 supercomputer at the Oak Ridge National Laboratory Leadership Computing Facility with 1-6 NVIDIA Volta V100 GPUs per computer node and (2) an in-house Intel Xeon computer with 1-2 NVIDIA Tesla P100 GPUs. The performance and parallel scalability were measured for three molecular models of 1-, 2-, and 3-dimensional chemical systems, represented by carbon nanotubes, covalent organic frameworks, and water clusters.

Importance of Nuclear Quantum Effects on Aqueous Electrolyte Transport under Confinement in Ti3C2 MXenes.

Protons display a high chemical activity and strongly affect the charge storage capability in confined interlayer spaces of two-dimensional (2D) materials. As such, an accurate representation of proton dynamics under confinement is important for understanding and predicting charge storage dynamics in these materials. While often ignored in atomistic-scale simulations, nuclear quantum effects (NQEs), e.g., tunneling, can be significant under confinement even at room temperature. Using the thermostatted ring polymer molecular dynamics implementation of path integral molecular dynamics (PIMD) in conjunction with the ReaxFF force field, density functional tight binding (DFTB), and NequIP neural network potential simulations, we investigate the role of NQEs on proton and water transport in bulk water and aqueous electrolytes under confinement in Ti3C2 MXenes. Although overall NQEs are relatively small, especially in bulk, we find that they can alter both quantitative values and qualitative trends on both proton transport and water self-diffusion under confinement relative to classical MD predictions. Therefore, our results suggest the need for NQEs to be considered to simulate aqueous systems under confinement for both qualitative and quantitative accuracy.

Quantum chemical investigation of the effect of alkali metal ions on the dynamic structure of water in aqueous solutions.

We report quantum chemical molecular dynamics (MD) simulations based on the density-functional tight-binding (DFTB) method to investigate the effect of K+, Na+, and Mg2+ ions in aqueous solutions on the static and dynamic structure of bulk water at room temperature and with various concentrations. The DFTB/MD simulations were validated for the description of ion solvation in aqueous ionic solutions by comparing static pair distribution functions (PDFs) as well as the cation solvation shell between experimental and available ab initio DFT data. The effect of the cations on the water structure, as well as relative differences between K+, Na+, and Mg2+ cations, were analyzed in terms of atomically resolved PDFs as well as time-dependent Van Hove correlation functions (VHFs). The investigation of the VHFs reveals that salt ions generally slow down the dynamic decay of the pair correlations in the water solvation sphere, irrespective of the cation size or charge. The analysis of partial metal-oxygen VHFs indicates that there are long-lived correlations between water and Na+ over long distances, in contrast to K+ and Mg2+.

Structural and functional characterization of NEMO cleavage by SARS-CoV-2 3CLpro.

In addition to its essential role in viral polyprotein processing, the SARS-CoV-2 3C-like protease (3CLpro) can cleave human immune signaling proteins, like NF-κB Essential Modulator (NEMO) and deregulate the host immune response. Here, in vitro assays show that SARS-CoV-2 3CLpro cleaves NEMO with fine-tuned efficiency. Analysis of the 2.50 Å resolution crystal structure of 3CLpro C145S bound to NEMO226-234 reveals subsites that tolerate a range of viral and host substrates through main chain hydrogen bonds while also enforcing specificity using side chain hydrogen bonds and hydrophobic contacts. Machine learning- and physics-based computational methods predict that variation in key binding residues of 3CLpro-NEMO helps explain the high fitness of SARS-CoV-2 in humans. We posit that cleavage of NEMO is an important piece of information to be accounted for, in the pathology of COVID-19.

Accurate Free Energies for Complex Condensed-Phase Reactions Using an Artificial Neural Network Corrected DFTB/MM Methodology.

Semiempirical methods like density functional tight-binding (DFTB) allow extensive phase space sampling, making it possible to generate free energy surfaces of complex reactions in condensed-phase environments. Such a high efficiency often comes at the cost of reduced accuracy, which may be improved by developing a specific reaction parametrization (SRP) for the particular molecular system. Thiol-disulfide exchange is a nucleophilic substitution reaction that occurs in a large class of proteins. Its proper description requires a high-level ab initio method, while DFT-GAA and hybrid functionals were shown to be inadequate, and so is DFTB due to its DFT-GGA descent. We develop an SRP for thiol-disulfide exchange based on an artificial neural network (ANN) implementation in the DFTB+ software and compare its performance to that of a standard SRP approach applied to DFTB. As an application, we use both new DFTB-SRP as components of a QM/MM scheme to investigate thiol-disulfide exchange in two molecular complexes: a solvated model system and a blood protein. Demonstrating the strengths of the methodology, highly accurate free energy surfaces are generated at a low cost, as the augmentation of DFTB with an ANN only adds a small computational overhead.

Structural and functional characterization of NEMO cleavage by SARS-CoV-2 3CLpro.

In addition to its essential role in viral polyprotein processing, the SARS-CoV-2 3C-like (3CLpro) protease can cleave human immune signaling proteins, like NF-κB Essential Modulator (NEMO) and deregulate the host immune response. Here, in vitro assays show that SARS-CoV-2 3CLpro cleaves NEMO with fine-tuned efficiency. Analysis of the 2.14 Å resolution crystal structure of 3CLpro C145S bound to NEMO 226-235 reveals subsites that tolerate a range of viral and host substrates through main chain hydrogen bonds while also enforcing specificity using side chain hydrogen bonds and hydrophobic contacts. Machine learning- and physics-based computational methods predict that variation in key binding residues of 3CLpro- NEMO helps explain the high fitness of SARS-CoV-2 in humans. We posit that cleavage of NEMO is an important piece of information to be accounted for in the pathology of COVID-19.

Design of tough adhesive from commodity thermoplastics through dynamic crosslinking.

Tough adhesives provide resistance against high debonding forces, and these adhesives are difficult to design because of the simultaneous requirement of strength and ductility. Here, we report a design of tough reversible/recyclable adhesive materials enabled by incorporating dynamic covalent bonds of boronic ester into commodity triblock thermoplastic elastomers that reversibly bind with various fillers and substrates. The spectroscopic measurements and density functional theory calculations unveil versatile dynamic covalent binding of boronic ester with various hydroxy-terminated surfaces such as silica nanoparticles, aluminum, steel, and glass. The designed multiphase material exhibits exceptionally high adhesion strength and work of debonding with a rebonding capability, as well as outstanding mechanical, thermal, and chemical resistance properties. Bonding and debonding at the interfaces dictate hybrid material properties, and this revelation of tailored dynamic interactions with multiple interfaces will open up a new design of adhesives and hybrid materials.

Investigating the Accuracy of Water Models through the Van Hove Correlation Function.

We present molecular-simulation-based calculations of the Van Hove correlation function (VHF) of water using multiple modeling approaches: classical molecular dynamics with simple three-site nonpolarizable models, with a polarizable model, and with a reactive force field; density functional tight-binding molecular dynamics; and ab initio molecular dynamics. Due to the many orders of magnitude difference in the computational cost of these approaches, we investigate how small and short the simulations can be while still yielding sufficiently accurate and interpretable results for the VHF. We investigate the accuracy of the different models by comparing them to recently published inelastic X-ray scattering measurements of the VHF. We find that all of the models exhibit qualitative agreement with the experiments, and in some models and for some properties, the agreement is quantitative. This work lays the foundation for future simulation approaches to calculating the VHF for aqueous solutions in bulk and under nanoconfinement.

Development of Density-Functional Tight-Binding Parameters for the Molecular Dynamics Simulation of Zirconia, Yttria, and Yttria-Stabilized Zirconia.

In this work, a set of density-functional tight-binding (DFTB) parameters for the Zr-Zr, Zr-O, Y-Y, Y-O, and Zr-Y interactions was developed for bulk and surface simulations of ZrO2 (zirconia), Y2O3 (yttria), and yttria-stabilized zirconia (YSZ) materials. The parameterization lays the ground work for realistic simulations of zirconia-, yttria-, and YSZ-based electrolytes in solid oxide fuel cells and YSZ-based catalysts on long timescales and relevant size scales. The parameterization was validated for the zirconia and yttria polymorphs observed under standard conditions based on density functional theory calculations and experimental data. Additionally, we performed DFTB-based molecular dynamics (MD) simulations to compute structural and vibrational properties of these materials. The results show that the parameters can give a qualitatively correct phase ordering of zirconia, where the tetragonal phase is more stable than the cubic phase at a lower temperature. The lattice parameters are only slightly overestimated by 0.05-0.1 Å (2% error), still within the typical accuracy of first-principles methods. Additionally, the MD results confirm that zirconia and yttria phases are stable against transformations under standard conditions. The parameterization also predicts that vibrational spectra are within the range of 100-1000 cm-1 for zirconia and 100-800 cm-1 for yttria, which is in good agreement with predictions both from full quantum mechanics and a recently developed classical force field. To further demonstrate the advantage of the developed DFTB parameters in terms of computational resources, we conducted DFTB/MD simulations of the YSZ4 and YS12 models containing approximately 750 atoms.

Methane Adsorption on Heteroatom-Modified Maquettes of Porous Carbon Surfaces.

Experimental and theoretical studies disagree on the energetics of methane adsorption on carbon materials. However, this information is critical for the rational design and optimization of the structure and composition of adsorbents for natural gas storage. The delicate nature of dispersion interactions, polarization of both the adsorbent and the adsorbate, interplay between H-bonding and tetrel bonding, and induced dipole/Coulomb interactions inherent to methane physisorption require computational treatment at the highest possible level of theory. In this study, we employed the smallest reasonable computational model, a maquette of porous carbon surfaces with a central site for substitution and methane binding. The most accurate predictions of methane adsorption energetics were achieved by electron-correlated molecular orbital theory CCSD(T) and hybrid density functional theory MN15 calculations employing a saturated, all-electron basis set. The characteristic geometry of methane adsorption on a carbon surface ("lander approach") arises due to bonding interactions of the adsorbent π-system with the proximal H-C bonds of methane, in addition to tetrel bonding between the antibonding orbital of the distal C-H bond and the central atom of the maquette (C, B, or N). The polarization of the electron density, structural deformations, and the comprehensive energetic analysis clearly indicate a ∼3 kJ mol-1 preference for methane binding on the N-substituted maquette. The B-substituted maquette showed a comparable or lower binding energy than the unsubstituted, pure C model, depending on the level of theory employed. The calculated thermodynamic results indicate a strategy for incorporating electron-enriched substitutions (e.g., N) into carbon materials as a way to increase methane storage capacity over electron-deficient (e.g., B) modifications. The thermochemical analysis was revised for establishing a conceptual agreement between the experimental isosteric heat of adsorption and the binding enthalpies from statistical thermodynamics principles.

Encapsulation of Aromatic Guests in the Bisporphyrin Cavity of a Double-Stranded Spiroborate Helicate: Thermodynamic and Kinetic Studies and the Encapsulation Mechanism.

A double-stranded spiroborate helicate bearing a bisporphyrin unit in the middle forms an inclusion complex with electron-deficient aromatic guests that are sandwiched between the porphyrins. In the present study, we systematically investigated the effects of size, electron density, and substituents of a series of aromatic guests on inclusion complex formations within the bisporphyrin. The thermodynamic and kinetic behaviors during the guest-encapsulation process were also investigated in detail. The guest-encapsulation abilities in the helicate increased with the increasing core sizes of the electron-deficient aromatic guests and decreased with the increasing bulkiness and number of substituents of the guests. Among the naphthalenediimide derivatives, those with bulky N-substituents at both ends hardly formed an inclusion complex. Instead, they formed a [2]rotaxane-like inclusion complex through the water-mediated dynamic B-O bond cleavage/reformation of the spiroborate groups of the helicate, which enhanced the conformational flexibility of the helicate to enlarge the bisporphyrin cavity and form an inclusion complex. Based on the X-ray crystal structure of a unique pacman-like 1:1 inclusion complex between the helicate and an ammonium cation as well as the molecular dynamics simulation results, a plausible mechanism for the inclusion of a planar aromatic guest within the helicate is also proposed.