Deep reinforcement learning in chemistry: A review.

Reinforcement learning (RL) has been applied to various domains in computational chemistry and has found wide-spread success. In this review, we first motivate the application of RL to chemistry and list some broad application domains, for example, molecule generation, geometry optimization, and retrosynthetic pathway search. We set up some of the formalism associated with reinforcement learning that should help the reader translate their chemistry problems into a form where RL can be used to solve them. We then discuss the solution formulations and algorithms proposed in recent literature for these problems, the advantages of one over the other, together with the necessary details of the RL algorithms they employ. This article should help the reader understand the state of RL applications in chemistry, learn about some relevant actively-researched open problems, gain insight into how RL can be used to approach them and hopefully inspire innovative RL applications in Chemistry.

DFT Studies of Dimethylaminophenyl-Substituted Phthalocyanine and Its Silver Complexes.

The dimethylaminophenyl-substituted silver phthalocyanine [dmaphPcAg] can be used as a UV-vis photoinitiator for in situ preparation of a silver/polymer nanocomposite. To verify early steps of the supposed mechanism of radical polymerization, we performed quantum chemical calculations of m[dmaphPcAg]q complexes with charges q = +1 to -2 in the two lowest spin states m, of a free ligand and its dehydrogenated/deprotonated products m[dmaphPcHn]q, n = 2 to 0, q = 0, -1 or -2, in the lowest spin states m. The calculated electronic structures and electron transitions of all the optimized structures in CHCl3 solutions are compared with experimental EPR and UV-vis spectra, respectively. The unstable 3[dmaphPcAg]+ species deduced only from previous EPR spin trap experiments was identified. In addition to 2[dmaphPcAg]0, our results suggest the coexistence of both reaction intermediates 1[dmaphPcAg]- and 3[dmaphPcAg]- in reaction solutions. Silver nanoparticle formation is a weak point of the supposed reaction mechanism from the energetic, stereochemistry, and electronic structure points of view.

Influence of Ancillary Ligand on the Photosensitization and Photophysical Properties of Red Luminescent Tri-fluorinated Β-diketonate Eu(III) Complexes.

Highly emissive ternary Eu(III) complexes were synthesized with a tri-fluorinated ß-diketone as principal ligand and heterocyclic aromatic compounds as ancillary ligands to assess their utility as an illuminating material for display devices and other optoelectronics. The general characterizations, regarding the coordinating facets of complexes were accomplished via various spectroscopic techniques. Thermal stability was investigated via TGA/DTA. Photophysical analysis was accomplished by PL studies, Band gap value, color parameters and J-O analysis. DFT calculations were performed adopting geometrically optimized structure of complexes. Superb thermal stability has been achieved in complexes, which decides their concrete candidature for display devices. The bright red luminescence of complexes is ascribed to 5D0 → 7F2 transition of Eu(III) ion. Colorimetric parameters unlocked the applicability of complexes as warm light source and J-O parameters adequately summarized the coordinating surrounding around the metal ion. Various radiative properties were also evaluated which suggested the prospective use of complexes in lasers and other optoelectronic devices. The band gap and Urbach band tail, procured from absorption spectra, revealed the semiconducting behavior of synthesized complexes. DFT studies rendered the energies of FMO and various other molecular parameters. It can be summarized from the photophysical and optical analysis of synthesized complexes that these complexes are virtuous luminescent materials and possess potentiality to be used in diverse domain of display devices.

Computational investigation of pyrazinamide drugs and its transition metal complexes using a DFT approach.

Pyrazinamide, an antituberculosis but documented toxic drug, is subjected to computational investigation along with the metal complexes via a DFT approach to predict the structure-activity and structure-toxicity relationship. 6-31G(d,p) basis set was used for Zn, Ni, Mn, Fe, and Co, while the SDD basis set was applied to Cu, Cr, Cd, and Hg. Several reactivity parameters and charge distribution were calculated and the reactivity profile was estimated. The complexes were found to be soft and polarizable which could be responsible for their binding with bacterial targets to inhibit their growth. In contrast, pyrazinamide which is found to be hard among all is susceptible to being toxic. Moreover, the electronegative nature of the complexes can endow them with a better antibacterial effect. Since metal complexes have been found to be less toxic and more biologically interactive by computational methods, they can be employed as potent drugs for the cure of tuberculosis.

A DFT Study of Solvent and Substituent Effects on the Adsorptive and Photovoltaic Properties of Some Selected Porphyrin Derivatives for DSSC Application.

A DFT/TD-DFT method was employed to study the effects of structural modification and solvent variation on the solubility, adsorptive, and photovoltaic properties of six porphyrins (A-F) obtained by structurally modifying two literature porphyrins A and D. The properties of interest were studied in vacuum, acetonitrile (AcCN), dichloromethane (DCM), dimethyl sulphoxide (DMSO), and ethanol (EtOH) for possible application of the molecules as sensitizers in dye-sensitized solar cells (DSSCs). Electronic absorption properties of the molecules were computed via potential energy surface scan, and thermodynamic data were obtained by DFT calculations in the selected media. Solubility properties of the molecules were mostly enhanced with DMSO as the solvent. The adsorptivity of the molecules onto mesoporous titanium (IV) oxide surface were predicted to be enhanced in the presence of DMSO. Most of the molecules were found to exhibit their highest photovoltaic activity measured in terms of the incident photon conversion efficiency (IPCE) in AcCN and DCM, rather than in DMSO due to its high viscosity and the ability to use its oxygen to form the catenating O-Ti4+ bond with the Ti4+ of the TiO2, causing inhibition of electron movement on the semiconductor surface. In general, the computed photovoltaic (PV) properties were found to be enhanced with -CO2H group as the substituent, and in AcCN or DCM as the solvent.

In-toto non-destructive assay methodology for the elimination of metallic waste from particle accelerators after melting.

Melting of metallic waste reduces the waste volume, allows more accurate radiological characterization, and minimizes handling at the waste production site. This paper proposes a new non-destructive assay methodology to radiologically characterize low- and intermediate-level (LILW) waste before melting. A non-destructive assay technique is developed and qualified using geometry optimization technique and sample analysis after melting. Additionally, we present an operational methodology to predict the activity values of the major gamma emitters based on the average dose rate measurements.

Combining biophysical models and machine learning to optimize implant geometry and stimulation protocol for intraneural electrodes.

Objective.Peripheral nerve interfaces have the potential to restore sensory, motor, and visceral functions. In particular, intraneural interfaces allow targeting deep neural structures with high selectivity, even if their performance strongly depends upon the implantation procedure and the subject's anatomy. Currently, few alternatives exist for the determination of the target subject structural and functional anatomy, and statistical characterizations from cadaveric samples are limited because of their high cost. We propose an optimization workflow that can guide both the pre-surgical planning and the determination of maximally selective multisite stimulation protocols for implants consisting of several intraneural electrodes, and we characterize its performance in silico. We show that the availability of structural and functional information leads to very high performances and allows taking informed decisions on neuroprosthetic design.Approach.We employ hybrid models (HMs) of neuromodulation in conjunction with a machine learning-based surrogate model to determine fiber activation under electrical stimulation, and two steps of optimization through particle swarm optimization to optimize in silico implant geometry, implantation and stimulation protocols using morphological data from the human median nerve at a reduced computational cost.Main results.Our method allows establishing the optimal geometry of multi-electrode transverse intra-fascicular multichannel electrode implants, the optimal number of electrodes to implant, their optimal insertion, and a set of multipolar stimulation protocols that lead in silico to selective activation of all the muscles innervated by the human median nerve.Significance.We show how to use effectively HMs for optimizing personalized neuroprostheses for motor function restoration. We provide in-silico evidences about the potential of multipolar stimulation to increase greatly selectivity. We also show that the knowledge of structural and functional anatomies of the target subject leads to very high selectivity and motivate the development of methods for theirin vivocharacterization.

Strain data augmentation enables machine learning of inorganic crystal geometry optimization.

Machine-learning (ML) models offer the potential to rapidly evaluate the vast inorganic crystalline materials space to efficiently find materials with properties that meet the challenges of our time. Current ML models require optimized equilibrium structures to attain accurate predictions of formation energies. However, equilibrium structures are generally not known for new materials and must be obtained through computationally expensive optimization, bottlenecking ML-based material screening. A computationally efficient structure optimizer is therefore highly desirable. In this work, we present an ML model capable of predicting the crystal energy response to global strain by using available elasticity data to augment the dataset. The addition of global strains improves our model's understanding of local strains too, significantly improving the accuracy of energy predictions on distorted structures. This allows us to construct an ML-based geometry optimizer, which we used for improving the predictions of formation energy for structures with perturbed atomic positions.

Automated handling of complex chemical structures in Z-matrix coordinates-The chemcoord library.

In this work, we present a fully automated method for the construction of chemically meaningful sets of hierarchical nonredundant internal coordinates (ICs; also commonly denoted as Z-matrices) from the Cartesian coordinates of a molecular system. Particular focus is placed on avoiding ill-definitions of angles and dihedrals due to linear arrangements of atoms, to consistently guarantee a well-defined transformation to Cartesian coordinates, even after structural changes. The representations thus obtained are particularly well suited for pathway construction in double-ended methods for transition state search and optimizations with nonlinear constraints. Analytical gradients for the transformation between the coordinate systems were derived for analytical geometry optimizations purely in Z-matrix coordinates. The geometry optimization was coupled with a Symbolic Algebra package to support arbitrary nonlinear constraints in Z-matrix coordinates, while retaining analytical energy gradient conversion. The difference to the commonly used nonhierarchical IC transformations is discussed. Sample applications are provided for a number of common chemical reactions and illustrative examples.

Theoretical foundation for designing multilayer Halbach array magnets for benchtop NMR and MRI.

Multilayer Halbach array magnets support portable NMR and MRI, but optimizing their design to maximize performance and minimize the use of expensive magnet materials is challenging. This is partly because our theoretical understanding of such arrays is incomplete and computationally intensive. Here we provide a theoretical description of the magnetic field distribution and we demonstrate that inhomogeneity is greatest along the z axis in multilayer Halbach array magnets. This allows the configuration of the multilayer Halbach array magnets to be optimized in a way that takes into account homogeneity, magnet volume, and magnetic flux density. At the same time, our description simplifies the design of multilayer array magnets, while accommodating the possibility of different outer radii, lengths for each layer array, or the presence of separation between the rings. We validated the theoretical description in simulations of a three-layer Halbach array magnet, then with a prototype three-layer 1-T Halbach array magnet. After adjusting the position of magnet blocks in the neighboring rings, we achieved homogeneity of 220 ppm for a standard 5 mm NMR tube while the inner diameter of the magnet is 20 mm. Our work provides a theoretical foundation for designing multilayer Halbach array magnets to maximize homogeneity and minimize the use of magnet materials.

Virtual Screening and Quantum Chemistry Analysis for SARS-CoV-2 RNA-Dependent RNA Polymerase Using the ChEMBL Database: Reproduction of the Remdesivir-RTP and Favipiravir-RTP Binding Modes Obtained from Cryo-EM Experiments with High Binding Affinity.

The novel coronavirus, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), was identified as the pathogenic cause of coronavirus disease 2019 (COVID-19). The RNA-dependent RNA polymerase (RdRp) of SARS-CoV-2 is a potential target for the treatment of COVID-19. An RdRp complex:dsRNA structure suitable for docking simulations was prepared using a cryo-electron microscopy (cryo-EM) structure (PDB ID: 7AAP; resolution, 2.60 Å) that was reported recently. Structural refinement was performed using energy calculations. Structure-based virtual screening was performed using the ChEMBL database. Through 1,838,257 screenings, 249 drugs (37 approved, 93 clinical, and 119 preclinical drugs) were predicted to exhibit a high binding affinity for the RdRp complex:dsRNA. Nine nucleoside triphosphate analogs with anti-viral activity were included among these hit drugs, and among them, remdesivir-ribonucleoside triphosphate and favipiravir-ribonucleoside triphosphate adopted a similar docking mode as that observed in the cryo-EM structure. Additional docking simulations for the predicted compounds with high binding affinity for the RdRp complex:dsRNA suggested that 184 bioactive compounds could be anti-SARS-CoV-2 drug candidates. The hit bioactive compounds mainly consisted of a typical noncovalent major groove binder for dsRNA. Three-layer ONIOM (MP2/6-31G:AM1:AMBER) geometry optimization calculations and frequency analyses (MP2/6-31G:AMBER) were performed to estimate the binding free energy of a representative bioactive compound obtained from the docking simulation, and the fragment molecular orbital calculation at the MP2/6-31G level of theory was subsequently performed for analyzing the detailed interactions. The procedure used in this study represents a possible strategy for discovering anti-SARS-CoV-2 drugs from drug libraries that could significantly shorten the clinical development period for drug repositioning.

Geometry Optimization of the Runner Insert for Improving the Appearance Quality of the Injection Molded Auto Part.

Appearance quality is one of the most important indexes for many injection-molded products, like optical parts, automotive parts etc., especially at the area near the injection gate. Different from the work that focused on designing the dimensions of the runner, this work proposed a method which is based on the insert technology to improve the appearance quality of a standard automotive part. An insert was introduced into the runner system which located before the gate. Three different shapes of this insert (circular, rectangular and diamond) were used to study the effect of geometrical factors on the appearance quality in this paper. All inserts were parameterized to describe their location and dimensions. Based on the geometrical design parameters, expected improvement optimization problem about the appearance quality were solved by using sequential approximate optimization method. The appearance qualities of three cases are improved by 13.77%, 21.56%, 14.37% respectively. Results showed that the best geometrical design scheme of the insert is rectangular with the optimal geometrical location and dimensions. The reasons were discussed by investigating the flow and thermal history in detail. Compared with the design case without any insert, the heat was absorbed and the velocity field was changed by the insert before the polymer melts ran into the cavity. It changed the complicated thermo-mechanical history inside the material during the entire processing history, which improved the final appearance quality of this auto part.

A Theoretical Study on the Medicinal Properties and Eletronic Structures of Platinum(IV) Anticancer Agents With Cl Substituents.

In this paper, we selected Pt(en)Cl4, Pt(dach)Cl4, and Pt(bipy)Cl4 with gradually increasing ligands to explore the ligand effect on the properties of platinum(IV) anticancer drugs. The electronic structures and multiple drug properties of these three complexes were studied at the LSDA/SDD level using the density functional theory (DFT) method. By comparing the gap between the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO), electron affinity, atomic charge population, and natural bond orbital (NBO), we found that the order of reducibility is Pt(bipy)Cl4 > Pt(en)Cl4 > Pt(dach)Cl4. Our research can provide the theoretical basis for the development of anticancer drugs.

Rigorous mathematical optimization of synthetic hepatic vascular trees.

In this paper, we introduce a new framework for generating synthetic vascular trees, based on rigorous model-based mathematical optimization. Our main contribution is the reformulation of finding the optimal global tree geometry into a nonlinear optimization problem (NLP). This rigorous mathematical formulation accommodates efficient solution algorithms such as the interior point method and allows us to easily change boundary conditions and constraints applied to the tree. Moreover, it creates trifurcations in addition to bifurcations. A second contribution is the addition of an optimization stage for the tree topology. Here, we combine constrained constructive optimization (CCO) with a heuristic approach to search among possible tree topologies. We combine the NLP formulation and the topology optimization into a single algorithmic approach. Finally, we attempt the validation of our new model-based optimization framework using a detailed corrosion cast of a human liver, which allows a quantitative comparison of the synthetic tree structure with the tree structure determined experimentally down to the fifth generation. The results show that our new framework is capable of generating asymmetric synthetic trees that match the available physiological corrosion cast data better than trees generated by the standard CCO approach.

Improving Localization Accuracy under Constrained Regions in Wireless Sensor Networks through Geometry Optimization.

In addition to various estimation algorithms, the target localization accuracy in wireless sensor networks (WSNs) can also be improved from the perspective of geometry optimization. Note that existing placement strategies are mainly aimed at unconstrained deployment regions, i.e., the positions of sensors are arbitrary. In this paper, considering factors such as terrain, communication, and security, the optimal range-based sensor geometries under circular deployment region and minimum safety distance constraints are proposed. The geometry optimization problem is modeled as a constrained optimization problem, with a D-optimality-based (maximizing the determinant of FIM matrix) scalar function as the objective function and the irregular feasible deployment regions as the constraints. We transform the constrained optimization problem into an equivalent form using the introduced maximum feasible angle and separation angle, and discuss the optimal geometries based on the relationship between the minimum safety distance and the maximum feasible angle. We first consider optimal geometries for two and three sensors in the localization system, and then use their findings to extend the study to scenarios with arbitrary numbers of sensors and arbitrarily shaped feasible regions. Numerical simulation results are included to verify the theoretical conclusions.

A second solvatomorph of poly[[µ4-N,N'-(1,3,5-oxadiazinane-3,5-di-yl)bis-(carbamoyl-methano-ato)]nickel(II)dipotassium]: crystal structure, Hirshfeld surface analysis and semi-empirical geometry optimization.

The title compound, poly[tri-aqua-bis-[µ4-N,N'-(1,3,5-oxadiazinane-3,5-di-yl)bis(carbamoyl-methano-ato)]dinickel(II)tetra-potassium], [K4Ni2(C7H6N4O7)2(H2O)3] n , is a second solvatomorph of poly[(µ4-N,N'-(1,3,5-oxadiazinane-3,5-di-yl)bis-(carbamoyl-methano-ato)nickel(II)dipotassium] reported previously [Plutenko et al. (2021 ▸). Acta Cryst. E77, 298-304]. The asymmetric unit of the title compound includes two structurally independent complex anions [Ni(C7H6N4O7)]2-, which exhibit an L-shaped geometry and consist of two almost flat fragments perpendicular to one another: the 1,3,5-oxadiazinane fragment and the fragment including other atoms of the anion. The central Ni atom is in a square-planar N2O2 coordination arrangement formed by two amide N and two carboxyl-ate O atoms. In the crystal, the title compound forms a layered structure in which layers of negatively charged complex anions and positively charged potassium cations are stacked along the a-axis direction. The polymeric framework is stabilized by a system of hydrogen-bonding inter-actions in which the water mol-ecules act as donors and the carb-oxy-lic, amide and water O atoms act as acceptors.

Controlling Resonator Nonlinearities and Modes through Geometry Optimization.

Controlling the nonlinearities of MEMS resonators is critical for their successful implementation in a wide range of sensing, signal conditioning, and filtering applications. Here, we utilize a passive technique based on geometry optimization to control the nonlinearities and the dynamical response of MEMS resonators. Also, we explored active technique i.e., tuning the axial stress of the resonator. To achieve this, we propose a new hybrid shape combining a straight and initially curved microbeam. The Galerkin method is employed to solve the beam equation and study the effect of the different design parameters on the ratios of the frequencies and the nonlinearities of the structure. We show by adequately selecting the parameters of the structure; we can realize systems with strong quadratic or cubic effective nonlinearities. Also, we investigate the resonator shape effect on symmetry breaking and study different linear coupling phenomena: crossing, veering, and mode hybridization. We demonstrate the possibility of tuning the frequencies of the different modes of vibrations to achieve commensurate ratios necessary for activating internal resonance. The proposed method is simple in principle, easy to fabricate, and offers a wide range of controllability on the sensor nonlinearities and response.

Distance-guided protein folding based on generalized descent direction.

Advances in the prediction of the inter-residue distance for a protein sequence have increased the accuracy to predict the correct folds of proteins with distance information. Here, we propose a distance-guided protein folding algorithm based on generalized descent direction, named GDDfold, which achieves effective structural perturbation and potential minimization in two stages. In the global stage, random-based direction is designed using evolutionary knowledge, which guides conformation population to cross potential barriers and explore conformational space rapidly in a large range. In the local stage, locally rugged potential landscape can be explored with the aid of conjugate-based direction integrated into a specific search strategy, which can improve the exploitation ability. GDDfold is tested on 347 proteins of a benchmark set, 24 template-free modeling (FM) approaches targets of CASP13 and 20 FM targets of CASP14. Results show that GDDfold correctly folds [template modeling (TM) score ≥ = 0.5] 316 out of 347 proteins, where 65 proteins have TM scores that are greater than 0.8, and significantly outperforms Rosetta-dist (distance-assisted fragment assembly method) and L-BFGSfold (distance geometry optimization method). On CASP FM targets, GDDfold is comparable with five state-of-the-art full-version methods, namely, Quark, RaptorX, Rosetta, MULTICOM and trRosetta in the CASP 13 and 14 server groups.

Evaluation of density functional theory for a large and diverse set of organic and inorganic equilibrium structures.

Density functional theory (DFT) has been extensively benchmarked for energetic properties; however, less attention has been given to equilibrium structures and the effect of using a certain DFT geometry on subsequent energetic properties. We evaluate the performance of 52 contemporary DFT methods for obtaining the structures of 122 species in the W4-11-GEOM database. This dataset includes a total of 246 unique bonds: 117 H─X, 65 X─Y, 49 X═Y, and 15 XY bonds (where X and Y are first- and second-row atoms) and 133 key bond angles: 96 X-Y-H, 22 X-Y-Z, and 15 H-X-H angles. The reference geometries are optimized at the CCSD(T)/jul-cc-pV(n+d)Z level of theory (n = 5, 6). The performance of DFT is evaluated in conjunction with the Def2-nZVPP (n = T, Q), cc-pV(T+d)Z, and jul-cc-pV(T+d)Z basis sets. The root-mean-square deviations (RMSDs) over the bond distances of the best performing functionals from each rung of Jacob's Ladder are 0.0086 (SOGGA11), 0.0088 (τ-HCTH), 0.0059 (B3LYP), 0.0054 (TPSSh), and 0.0032 (DSD-PBEP86) Å. We evaluate the effect of the choice of the DFT geometry on subsequent molecular energies calculated with W1-F12 theory. Geometries obtained with GGA and MGGA methods result in large RMSDs in the subsequent W1-F12 energies; however, six hybrid GGA functionals (B3LYP, B3P86, mPW3PBE, B3PW91, mPW1LYP, and X3LYP) result in an excellent performance with RMSDs between 0.25 and 0.30 kJ mol-1 relative to the CCSD(T)/CBS reference geometries. The B2GP-PLYP and mPW2-PLYP DHDFT methods result in near-CCSD(T) accuracy with RMSDs of 0.11 and 0.10 kJ mol-1 , respectively.

Improving small molecule force fields by identifying and characterizing small molecules with inconsistent parameters.

Many molecular simulation methods use force fields to help model and simulate molecules and their behavior in various environments. Force fields are sets of functions and parameters used to calculate the potential energy of a chemical system as a function of the atomic coordinates. Despite the widespread use of force fields, their inadequacies are often thought to contribute to systematic errors in molecular simulations. Furthermore, different force fields tend to give varying results on the same systems with the same simulation settings. Here, we present a pipeline for comparing the geometries of small molecule conformers. We aimed to identify molecules or chemistries that are particularly informative for future force field development because they display inconsistencies between force fields. We applied our pipeline to a subset of the eMolecules database, and highlighted molecules that appear to be parameterized inconsistently across different force fields. We then identified over-represented functional groups in these molecule sets. The molecules and moieties identified by this pipeline may be particularly helpful for future force field parameterization.