Journal of Computational Chemistry

Bojarski, Krzysztof K.

doi: 10.1002/jcc.70161pmid: 40556556

Glycosaminoglycans (GAGs) are linear, negatively charged carbohydrates that modulate enzymatic activity in the extracellular matrix. Their high flexibility and specificity in protein‐GAG interactions pose challenges for both experimental and computational studies. Here, the repulsive scaling replica exchange molecular dynamics (RS‐REMD) method, combined with molecular mechanics generalized born surface area (MM‐GBSA), was implemented using the CHARMM36m force field to evaluate its ability to guide ligands to their native binding sites in seven protein‐GAG/carbohydrate complexes. A five machine learning (ML)‐based models including fully connected neural network (FCNN), linear regression, LightGBM, random forest and support vector regressor (SVR) were also trained to predict binding accuracy (RMSatd) based on MM‐GBSA energy components, protein‐GAG distances, and hydrogen bond counts. While MM‐GBSA values showed weak to moderate correlations with RMSatd, most of the trained AI models significantly improved the selection of native‐like binding poses with Random Forest model providing most accurate predictions. This study highlights the potential of integrating simulations with ML to refine molecular docking for flexible ligands like GAGs.

Scheiner, Steve

doi: 10.1002/jcc.70163pmid: 40528726

The interaction energies within noncovalent bonds can be partitioned into electrostatic, induction, and dispersive attractive elements. A set of complexes comprising halogen, chalcogen, pnicogen, and tetrel bonds, are studied by quantum chemical calculations to assess how each of these components can be understood on the basis of properties of the constituent monomers. The variation of the electrostatic term, which accounts for over half of the total attractive energy, can be approximated, but with only modest accuracy, by combination of the maximum and minimum of the electrostatic potential on the two subunits. Induction represents a smaller contribution to the total, but is well connected with the NBO interorbital transfer energy, as opposed to the reciprocal of the HOMO‐LUMO gap which behaves quite differently than IND. Of the various AIM parameters, both the bond critical point density and energy density are closely related to the full interaction energy.

Ursal, Kapil Dattatray; Sk, Md Fulbabu; Mahapatra, Subhasmita; Kar, Parimal

doi: 10.1002/jcc.70172pmid: 40583493

The intricate world of cellular processes relies significantly on the dual‐specificity tyrosine‐phosphorylation‐regulated kinase (DYRK) family of kinases, governing vital functions like brain development, splicing regulation, and apoptosis. DYRK1A, in particular, stands at the center of attention due to its pivotal role. Disruptions in its activity, whether through upregulation or downregulation, have profound implications, notably in neurological disorders and cancer progression. Understanding the impact of phosphorylation, a fundamental post‐translational modification, on DYRK1A is paramount. In this study, we delved into the complex interplay of phosphorylation and the effects of the abemaciclib inhibitor on DYRK1A conformational dynamics. We employed advanced techniques such as molecular dynamics simulations and the molecular mechanics Poisson‐Boltzmann surface area (MM/PBSA) scheme and deciphered the intricate dance of DYRK1A's structural elements during phosphorylation. Our exploration revealed intriguing details: the αC‐helix undergoing outward movement, a distorted αC‐helix, a wide‐open P‐loop, extended A‐loop, and role of electrostatic interactions shaping A‐loop dynamics. Notably, the interaction of specific residues, particularly Lys188, forming robust salt bridges with Asp307 and Glu203, plays a pivotal role in shaping the structure of the protein. Diving deeper, we conducted principal component analysis and conformational free energy sampling to uncover crucial structural intermediates. Moreover, our dynamic cross‐correlation map sheds light on the influence of phosphorylation by enhancing coordinated movements while dampening anti‐correlated motions across various domains. This nuanced understanding of DYRK1A kinase activation, driven by phosphorylation, not only enriches our knowledge but also holds promise in the development of targeted therapies for associated diseases.

Arra, Srilatha; Daidone, Isabella; Aschi, Massimiliano

doi: 10.1002/jcc.70144pmid: 40539257

The cluster‐in‐solvent approach, that is, the use of the quantum‐mechanical calculation of a spectral observable on a significant number of solute–solvent clusters extracted from semi‐classical simulations, is widely used in computational spectroscopy. However, identifying relevant coordinates for cluster selection remains a challenge. We previously developed the Ellipsoid Method for Cluster‐in‐Solvent (EMCS), an automated strategy for unbiased identification and statistical weighting of clusters. Yet, for larger solutes, EMCS can yield overly large solvent clusters that hinder conformational analysis. Here, we introduce a simple extension of EMCS that reduces cluster size, enabling its application to medium‐to‐large solutes. The method is validated through the computation of Vibrational Circular Dichroism (VCD) spectra, which are highly sensitive to solute–solvent interactions. Test cases include aqueous L‐alanine, aqueous dialanine, and (1S,2S)‐trans‐1‐amino‐2‐indanol in DMSO. For L‐alanine and trans‐1‐amino‐2‐indanol, computed spectra closely match experiment, with root‐mean‐square‐deviation (RMSD) values of 10.3 and 8.0, respectively, consistent with previous benchmarks. For aqueous dialanine, the main spectral features were reproduced, though discrepancies in the fine structure remain, likely due to limitations in capturing subtle solvation effects. Overall, the refined EMCS protocol enables efficient and non‐arbitrary sampling of solute–solvent clusters, offering a valuable tool for the structural analysis of solvation shells in complex molecular systems.

Wang, YiXu; Müller, Peter C.; Hemker, David; Dronskowski, Richard

doi: 10.1002/jcc.70167pmid: 40586640

The computer program LOPOSTER, available via GitHub, is introduced, capable of postprocessing the LOBSTER code results. LOPOSTER is designed to be particularly effective for analyzing large datasets with over 10,000 interactions and enormously reducing postprocessing time. LOPOSTER pioneers the automated processing of advanced bonding analysis results, including multicenter bonding, molecular‐orbital formation energy, and k‐dependent COHP, expanding the scope of routine chemical‐bonding investigations. In addition, LOPOSTER streamlines the postprocessing workflow by providing comprehensive results in a single execution, minimizing user intervention and potential errors. An example of chemical‐bonding analysis on NiNCN is provided, with visualization by LOPOSTER. LOPOSTER offers versatile analysis of interactions in NiNCN, enabling evaluations in real or reciprocal space, and based on atomic or molecular orbitals, catering to different analytic preferences. Various correlations between those interactions and magnetism in NiNCN are also explored. The electron‐rich features of an N=C=N π bond have been discussed from various perspectives.

Bonvicini, Andrea; Champagne, Benoît

doi: 10.1002/jcc.70149pmid: 40566831

We present pyHRSOA and pyTHSOA, two postprocessing Python codes for the computation of the circular differential scattering ratio, a dimensionless quantity which is the central observable for two novel and promising nonlinear chiroptical techniques: the hyper‐Rayleigh scattering optical activity (HRS‐OA) and the third‐harmonic scattering optical activity (THS‐OA) spectroscopies. From a computational point of view, the simulation of the HRS‐OA (THS‐OA) spectroscopy requires the calculations of five first (second) hyperpolarizabilities and these can be obtained from quadratic (cubic) responses functions by using the DALTON quantum chemistry software. However, the expressions for the chiral and achiral contributions to the scattered intensities are quite complicated because these contain a lot of terms. In particular, for HRS‐OA and THS‐OA 30 and 46 molecular invariants need to be computed, thus complicating their computational implementation. The postprocessing codes presented here can be used as black‐box tools for the simulations of HRS‐OA and THS‐OA spectroscopies. The source codes are available at https://gitlab.unamur.be/abonvici/pyhrsoa_pythsoa (DOI 10.5281/zenodo.15424494).

Sadiq, Zubair; Yang, Wenhong; Yang, Weisheng; Sun, Wen‐Hua

doi: 10.1002/jcc.70157pmid: 40526080

This study focuses on training a multi‐task learning (MTL) type machine learning (ML) model to predict diverse catalytic performance of 195 bis(imino)pyridine transition metal complexes toward ethylene polymerization, with comparison to their single‐task learning (STL) counterparts. The CatBoost MTL model outperforms all other models, showing predictions and generalization errors for the properties of catalytic activity (Rt2=0.741, R2 = 0.985, Q2 = 0.600), molecular weight (Rt2=0.873, R2 = 0.997, Q2 = 0.846), molecular weight distribution (Rt2=0.831, R2 = 0.999, Q2 = 0.839), and melting temperature (Rt2=0.813, R2 = 0.992, Q2 = 0.625) of the produced polymer. The interpretation of the model reveals that complexes with electron‐donating groups, simple alkyl groups (such as methyl groups etc.), and a higher degree of unsaturation (presence of double or triple bonds) positively influence the predicted properties. Subsequently, providing insights into the underlying mechanisms of variation in catalytic performance, new complexes are designed with superior catalytic performances.

Dutra, José Diogo Lisboa; Oliveira, Willyan Farias; Silva, Gustavo Santana; Bispo, Thiago Dias; Freire, Ricardo Oliveira

doi: 10.1002/jcc.70143pmid: 40546224

There is no doubt about the relevance that lanthanide‐based luminescent systems have gained in recent decades. These systems have applications in various technological fields, and the development of these materials has been of great scientific interest since the 1970s. We made the first version of the LUMPAC software available 10 years ago. In our publication, we highlighted the scarcity of studies concerning the use of theoretical computational methods to design new luminescent systems and to explain laboratory‐observed phenomena involving such systems. In this work, we present the LUMPAC 2.0 software. We show how the first version contributed to increasing the number of studies involving theoretical computational methods, and we evaluate its use since its launch. This second version introduces highly relevant features and implementations and will undoubtedly be an important tool in new studies that will be carried out over the next few years.

Ligorio, Raphael F.; Dos Santos, Leonardo H. R.; Krawczuk, Anna

doi: 10.1002/jcc.70158pmid: 40576460

Despite its name, the dipole interaction model (DIM) serves not only to adjust dipole moments due to atomic interactions but also to assess polarizabilities. Traditionally, polarizability calculations via DIM rely on matrix inversion, posing constraints on memory usage and computational time. Recent implementations have shown significant performance boosts by employing an iterative inversion solver, albeit reducing accuracy. In this paper, we present a direct approach for computing polarizabilities via iterative cycles, eliminating the need for matrix inversion. This allows for scaling up the model to hundreds of thousands of atoms without sacrificing precision, as often happens when simplifying the standard inversion procedure to reduce computational costs. Additionally, we have addressed memory issues associated with storing extensive arrays in standard implementations. Our advancement holds promise for diverse applications, providing an efficient method for exploring polarizabilities in various systems.

Tikhonov, Denis S.; Kalinin, Mikhail A.; Maryewski, Alexander A.; Avdoshin, Aleksandr A.; Dallakyan, Olgert; Vasilev, Nikita A.; Eliseev, Egor A.; Koch, Mandy; Rybkin, Vladimir V.; Artiukhin, Denis G.

doi: 10.1002/jcc.70148pmid: 40557783

A recent increase in targeted attacks using chemical warfare agents by dictators and authoritarian regimes against politicians, journalists, and other civilians is a major concern. To aid the civil investigators in identifying poisonous substances in such cases, we developed an algorithm and a lightweight and simple‐to‐use software, ToxicMassSceptic$$ \mathtt{ToxicMassSceptic} $$, with a database of 400 electron ionization mass spectra entries, which include many poisonous and explosive agents. The identification relies on a window‐based reduction of the experimental spectra and four statistical metrics that are combined into a single metametric. The software also features automatic spectral background removal. Furthermore, we provide the workflow for increasing the size of this database by performing theoretical calculations of mass spectra with a molecular dynamics‐based approach. The accuracy of both the theoretical prediction workflow and ToxicMassSceptic$$ \mathtt{ToxicMassSceptic} $$ is validated on the experimental spectra. Our results demonstrate that the proposed software package can aid in the preliminary identification of traces of poisonous and explosive substances.