In this work, density functional theory and natural bond orbital analysis are used to investigate the electronic structures of the [4Fe-4S] cluster in reduced high-potential iron–sulfur proteins. Calculated 3d-orbital occupancies of Fe atoms in the [4Fe-4S] cluster revealed a spin structure with two ferromagnetically coupled [2Fe-2S]⁺ subclusters that are antiferromagnetically coupled to each other. In addition, the chemical bonds between S–Fe were found to form primarily through the donation of electrons from S to Fe atoms.

Racemization of amino acid catalyzed by aldehyde and carboxylic acid has been proposed and widely used. Herein, we investigated the detailed racemization mechanism of alanine catalyzed by salicylaldehyde and acetic acid. Quantum chemistry calculations based on the density functional theory demonstrated that the dehydration reaction is the rate-determining step, and this dehydration step is enhanced by surrounding water molecules. We also discussed the dependence of the reaction rates on the type of amino acid.

Metalloproteins such as 4-hydroxyphenylpyruvate dioxygenase (HPPD) and Cytochrome P450 are important for agrochemical research. In classical mechanics, to reproduce the metal-coordination environment of metalloproteins, various models (Nonbonded model, Bonded model, Nonbonded/Bonded Hybrid model etc.) for metal complexes have been developed. In this study, we built force field parameters for the molecular dynamics simulation of HPPD and evaluated the Fe²⁺-coordination environment. The result revealed that the hybrid model is suitable for the simulation of HPPD and able to simulate the metal-ligand (water) exchange process.

Synthetic pathways of Keggin-type polyoxotungstates (POTs), which are expected to catalyze multi-electron transfer reactions, were investigated by first-principles calculations. The reaction pathways for the formation of new Keggin-type POTs from existing ones via ion exchange were explored by the Nudged Elastic Band (NEB) method, and the effects of the kind of heteroatom and the relative permittivity of the solvent were discussed.

π-stacked single-molecule junctions stacked with π-conjugated molecules have the potential to be used as building blocks for single-molecule scale three-dimensional integrated circuits. In this study, we investigate the relationship between π-stacking distance and conductance in face-to-face π-stacked single-molecule junctions with benzene as the monomer unit using the non-equilibrium Green's function (NEGF), which combines the Hückel molecular orbital (HMO) and density functional theory (DFT) methods. As the π-stack distance between two benzene molecules decreases, the pseudo-para junction, which is insulating, turns conductive. Furthermore, it was found that the cause of this change can be explained by orbital interactions.

In this study, we used a benchmark dataset to evaluate the impact of scaling with the extrapolation domain on the prediction performance of machine learning algorithms. We pseudo-divided the data into the interpolation domain (training data) and the extrapolation domain (test data) using a combination of UMAP (Uniform Manifold Approximation and Projection) and material domain knowledge. In anticipation of bridging interpolation and extrapolation domains in nonlinear machine learning algorithms, we evaluated how the scaling considering the extrapolation domain affects prediction performance in the extrapolation domain. For this evaluation, we used three nonlinear algorithms widely used in the MI (Materials Informatics) domain: XGB (XGBoost) regression, GP (Gaussian Process) regression, and SVR (Support Vector Regression). In this study, by defining the pseudo extrapolation domain, we established the approach for evaluating the prediction accuracy of machine learning models in the extrapolation domain, which is considered difficult to evaluate quantitatively. We also demonstrated that this method, which uses scaling that considers the extrapolation domain, is an effective method for improving prediction accuracy in the extrapolation domain while maintaining prediction accuracy in the interpolation domain.

We used molecular dynamics simulations to analyze the structural changes of medaka fish taste receptor proteins T1r2a-T1r3 and three ligands, L-glutamine, L-glutamic acid, and L-alanine, and the movement of water molecules around the ligands. L-glutamic acid, which had a low affinity in the experiment, showed many water molecules after the MD simulation, although the number of crystalline water molecules around the ligand was small in the crystal structure. We were also able to identify similar hydrogen bonds between residues and ligands that are conserved in other class C G protein-coupled receptors. In addition, several water molecules were observed to be alternately hydrogen-bonded to the side chain of the ligand with no characteristic lifetime.

The Xia–Bian–Kais (XBK) transformation, which maps the second quantized Hamiltonian into the Ising Hamiltonian, is a method for optimizing the configuration interaction (CI) wave function using an annealing computer. In this method, r replicas of qubit sequences, each of which corresponds to a Slater determinant, are used for mapping. To express a CI wave function with determinants having different magnitudes of coefficients, r should be increased. We developed a low-cost and highly-accurate ABCI (Annealing + Bayesian-optimization CI) method by introducing weight bits into the XBK method, which are determined by the Bayesian optimization.

We have been developing a method for coarse-graining the low-frequency vibration modes of molecular assemblies, which affords a numerical representation of the down-sized stiffness matrix. In this study, we present an analytical representation of the stiffness matrix based on the elastic-body modeling of molecular assemblies. Comparison between the numerical and analytical data allows the 13 parameters regarding the dimension and mechanical properties of the putative elastic body. The results for 57 molecular dimers with various hydrogen-bond multiplicity demonstrate that the obtained parameters were physically reasonable and well-reproduced the wavenumbers of normal-mode vibrations.

The values of the DPPH free radical scavenging ability values (IC₅₀) of ferulic acid and its derivatives have been estimated by machine learning with only molecular orbital (MO) energies as the explanatory variables. We use four machine learning regression methods, Partial Least Square (pls), Random Forest (rf), Multi-Layer Perceptron Neural Network with Optional Monotonicity Constraints (monmlp) and eXtreme Gradient Boosting with Linear model (xgbLinear), using R-package caret. We use 22 molecules for the training set and 6 molecules for the test set. The root mean square (RMS) errors of predicted values for the test set are used for estimating the precision of the training. The best result is obtained by xgbLinear just using two MO energies (HOMO and LUMO). It has been proved that the IC₅₀ values can be predicted by the molecular orbital energies only.

We have been developing the ABINIT-MP program for fragment molecular orbital (FMO) calculations over 20 years. Several improvements for accelerated processing were made after the release of Open Version 2 Revision 4 at September 2021. Functionalities were enhanced as well. In this short report, we summarize such developments toward the next release of Revision 8.

Thermodynamic properties and atomic configuration of PdRu alloy were investigated using Wang-Landau Monte Carlo method in combination with a newly-developed universal neural network potential. By using this new potential, excess energy of PdRu alloy was calculated. It is found that PdRu alloy in FCC lattice is unstable in the full range of alloy composition. This agrees with previous study based on density functional theory. The combined method was able to determine the configurational density of states, from which thermodynamic properties of the alloy were derived. It is found that when temperature increases, the excess free energy of the alloy is reduced, increasing the possibility of alloy mixing. Depending on the composition, transition peaks appear at finite temperatures where there are changes of preferable atomic arrangement due to the effect of temperature via the configurational entropy. In addition, the analyses on short-range order parameter and bond fraction show that PdRu alloy prefers to be in segregated form, where Pd and Ru are immiscible at low temperature, consistent with the experimental observations. The random mixing of Pd and Ru atoms in the form of solid-solution can occur at high temperature.

深層学習による反応予測は，人が識別パターンを設計する必要がないため最近注目されている．しかし有機遷移金属反応は一見複雑で，深層学習の適応例は殆どなかった．我々は，論文や特許などから集めた数万件の実験反応を，素反応に分解したデータベースを構築し，グラフニューラルネットワークを用いて学習させた．最高で97.1%の精度で反応が予測できた．

Multi-element alloy nanoparticles have attracted attention for their potentially high catalytic properties. However, a high degree of freedom in configurations of metal atoms within nanoparticle increases the distinct adsorption sites, making it difficult to theoretically analyze its catalytic properties because the first-principles calculation requires a considerable computational cost. In this study, we develop a sequential scheme to calculate hundreds of adsorption sites by employing a pre-trained universal neural network potential named PFP. Our automated scheme is applied to CO single-molecule adsorption of CO onto PtRuIr ternary alloy nanoparticles. The calculation results are first compared with DFT results to confirm the accuracy. Adsorption energies in the alloy systems are widely distributed in comparison with those of the monometal counterparts, indicating that the alloy nanoparticle includes adsorption sites with various catalytic activities.

A 3-dimensional domain swapping (3D-DS) phenomenon is a promising way to make a stable multimer because interactions in the multimer is about the same as that of a monomer; it would be useful to artificially design a functional protein's multimer. In this study, molecular dynamics (MD) simulation of cytochrome c(cyt c) that is known to form a 3D-DS dimer (PDB ID: 3NBS) was performed to study factors that enhance 3D-DS structure sampling. Our results show a difficulty of 3D-DS structure sampling and a necessity of a method such as the generalized ensemble method. The results also show the importance of a loop-flexibility to sample 3D-DS structure; when the bond and torsion angle potential coefficients are less than 10% of secondary structure, 3D-DS structures become easy to sample.