Search

In this paper, the effect of the ionic-controlling agent on dynamics behavior of the Cu–Ag system in ethylene glycol (EG) was studied for the first time via the molecular dynamic method. The optimized Cu–Ag bimetallic nanoparticles (BNPs) with different control agents are evaluated by dynamic stability, total energy, interaction energy, atom distribution of Cu–Ag–EG system and concentration distribution of different components. We found that the Ag atoms tend to be more stable on the outer surface of the Cu–Ag BNPs in the presence of NO3−, while for the presence of H+, on the contrary, Cu atoms tend to be located on the outer surface. Moreover, the interaction between Cu–Ag BNPs and control agents exhibits the following sequence: H+>SO42−>Cl−>Fe3+>Na+>NO3−. For the Cu–Ag–EG–NO3− system, the presence of a small amount of control agent makes the interfacial behavior of the Cu–Ag–EG system much more complicated. It indicates that the content of the ionic agent is not the more the better. On the base of maintaining the stability of the system, adding an appropriate amount of controlling agent can effectively control the structure of Cu–Ag BNPs, and then adjust the morphology and properties of the production. [ABSTRACT FROM AUTHOR]

The regularities of crystal structure organization were studied in a series of para- and ortho-halogenanilines using an approach based on comparison of interaction energies between molecules calculated by an ab initio method. The halogen substituent position in anilines significantly affects intermolecular interactions, differing between para- and ortho-halogen anilines. In para-halogen anilines, the amino group mainly acts as a hydrogen-bond donor, with no significant halogen bonds and weak stacking interactions. In ortho-halogen anilines, the amino group functions as both donor and acceptor, with stronger hydrogen and halogen bonding. Crystal packing analysis shows columnar organization across samples, with zig-zag columns in para-substituted anilines and triple columns in ortho-substituted ones. Overall, halogen bonds play a minor role, mainly connecting neighboring columns. [ABSTRACT FROM AUTHOR]

Four acridinium/acridine derivatives have been prepared by mechanochemical synthesis and their crystal structure, supramolecular framework, interaction energy calculation, thermal analysis, and photophysical properties are studied and presented in this manuscript. In the crystal structure, the one- and two-dimensional supramolecular framework is constructed via various strong and weak intermolecular interactions such as N‒H...Cl, O‒H...N, O‒H...Cl, C‒H...Cl, C‒H...O, C‒H...π, and π...π, respectively. The contribution of intermolecular interaction in the three-dimensional molecular packing is studied by the Hirshfeld surface analysis. The calculated total energy value of intermolecular interaction/contacts observed between the molecular pairs in the acridine compound is stronger than the energy value of intermolecular interaction/contacts of acridinium derivatives. The energy value of π...π contacts exhibited between the molecular pairs is significantly stronger than other weak interactions. The thermogravimetric analysis reveals that the acridinium derivatives degrade in three steps whereas the acridine compound undergoes a single-step degradation. Studies show that the acridinium derivatives exhibit a better photoluminescence quantum yield when compared to the acridine compound, and the acridine compound experiences a photoluminescence quenching due to charge transfer interactions. [ABSTRACT FROM AUTHOR]

To investigate the mechanism underlying the effect of mixed collector (MC) composed of diesel and span-80 on LRC flotation, a combination of experimental and molecular simulation approaches was employed. The flotation results demonstrated that the use of MC yielded superior flotation performance compared to diesel alone. The ratio of diesel to span-80 emerged as a key factor influencing the performance of MC collection. Furthermore, molecular dynamics (MD) simulations were employed to investigate the adsorption processes of dodecane and MC molecules on the LRC surface in the LRC/dodecane/water and LRC/MC/water systems, respectively. The analysis encompassed microstructure, dynamic properties, interaction energy, and hydrogen bonding. The findings indicated that the presence of span-80 molecules facilitated the direct entry of dodecane molecules in MC into "microscopic cavities," thereby enhancing the adsorption of MC on the LRC surface. The analysis of interaction energy revealed a stronger interaction between MC molecules and the coal surface, with van der Waals energy playing a major role in the adsorption process. Additionally, span-80 molecules induced hydrogen bonding adsorption on the LRC surface. [ABSTRACT FROM AUTHOR]

Understanding rare earth element (REE) binding to proteins enables the engineering of selective protein-based ligands for precise REE recovery. Lanmodulin (LanM), with notable REE selectivity and picomolar binding affinity, is a promising candidate. This study shows that LanM variants employ distinct inter-residue interactions for REE binding. We detail the thermodynamics and structural aspects of binding events in wild-type (WT) Methylorubrum extorquens LanM and five EF-hand residue variants (4P2A and 4D9X, X = N, A, H, M), using protein variant structure prediction, molecular dynamics simulations and binding motif exploration. We demonstrate strong agreement between experimental binding measurements (apparent Kd) and in silico binding energy scores of WT, 4 P2A, and 4D9X LanMs. We systematically investigate the role of solvent dielectric, sample multiple force fields, and initial protein structure bias on metal ion-binding energetics. In addition, we identify amino acids outside the direct metal binding motif crucial for coordinating the binding events which is corroborated with experimental binding characteristics of 4D9X variants. Computationally measured binding affinity with contribution from this secondary set of residues show agreement with the experimental Kd values and suggests how some point mutations can induce long-range structural perturbations to regulate metal ion-protein recognition and interactions. Finally, we analyze structural changes arising from alterations in side-chain flexibility of each amino acid on the protein backbone at the instant of metal binding and recognition – which manifests as altered helicity at a specific locus of the protein, a result that is corroborative of the observations from circular dichroism experiments.

We introduce the fragment‐pairwise Local Energy Decomposition (fp‐LED) scheme for precise quantification of individual interactions contributing to the binding energy of arbitrary chemical entities, such as protein‐ligand binding energies, lattice energies of molecular crystals, or association energies of large biomolecular assemblies. Using fp‐LED, we can assess whether the contribution to the binding energy arising from noncovalent interactions between pairs of molecular fragments in any chemical system is attractive or repulsive, and accurately quantify its magnitude at the coupled cluster level ‐ commonly considered as the “gold standard” of computational chemistry. Such insights are crucial for advancing molecular and material design strategies in fields like catalysis and therapeutic development. Illustrative applications across diverse fields demonstrate the versatility and accuracy of this theoretical framework, promising profound implications for fundamental understanding and practical applications. [ABSTRACT FROM AUTHOR]

This study aims to cast light on the nature of hydrogen bonds and their cooperativity that exists in alcohol-water ( $ \textrm{ROH}{({{\textrm{H}_2}\textrm{O}})_\textrm{n}} $ ROH ( H 2 O ) n ) ( $ \textrm{R} = - \textrm{C}{\textrm{H}_3}\textrm{ }({ - \textrm{Me}}),\textrm{ } - \textrm{C}{\textrm{H}_2}\textrm{C}{\textrm{H}_3}\textrm{ }({ - \textrm{Et}}) $ R = − C H 3 (− Me) , − C H 2 C H 3 (− Et)) (n = 1–10) clusters using atoms in molecules (AIM) theory and symmetry-adapted perturbation theory (SAPT). The investigation of optimum structures suggests that alcohol ( $ \textrm{ROH} $ ROH ) forms two hydrogen bonds, one primary hydrogen bond $ ({\textrm{O} - \textrm{H} \cdot \cdot \cdot \textrm{O}}) $ (O − H ⋅ ⋅ ⋅ O) , and one secondary hydrogen bond $ ({\textrm{C} - \textrm{H} \cdot \cdot \cdot \textrm{O}}) $ (C − H ⋅ ⋅ ⋅ O) Two types of hydrogen bonds are addressed in these studies, where alcohol works as the hydrogen bond donor $ ({\textrm{ROH} \cdot \cdot \cdot \textrm{O}{\textrm{H}_2}}) $ (ROH ⋅ ⋅ ⋅ O H 2 ) and hydrogen bond acceptor $ ({\textrm{RHO} \cdot \cdot \cdot \textrm{HOH}}) $ (RHO ⋅ ⋅ ⋅ HOH). The AIM results reveal that $ \textrm{RHO} \cdot \cdot \cdot \textrm{HOH} $ RHO ⋅ ⋅ ⋅ HOH and $ \textrm{ROH} \cdot \cdot \cdot \textrm{O}{\textrm{H}_2} $ ROH ⋅ ⋅ ⋅ O H 2 hydrogen bonds are partially covalent. The SAPT study demonstrates that in all the clusters studied herein, the electrostatic interaction between alcohol and water is the main attractive force, and its contribution may be two times larger than the separate corresponding contributions from the dispersion and the induction terms. The electron density (ρ) at the bond critical point (BCP) is a suitable parameter to assess the strength of the interaction. The current study shows the strength and nature of $ \textrm{ROH} \cdot \cdot \cdot \textrm{O}{\textrm{H}_2} $ ROH ⋅ ⋅ ⋅ O H 2 and $ \textrm{RHO} \cdot \cdot \cdot \textrm{HOH} $ RHO ⋅ ⋅ ⋅ HOH hydrogen bond, and cooperative effect in ( $ \textrm{ROH}{({{\textrm{H}_2}\textrm{O}})_\textrm{n}} $ ROH ( H 2 O ) n ) ( $ \textrm{R} = - \textrm{C}{\textrm{H}_3}\textrm{ }({ - \textrm{Me}}),\textrm{ } - \textrm{C}{\textrm{H}_2}\textrm{C}{\textrm{H}_3}\textrm{ }({ - \textrm{Et}}) $ R = − C H 3 (− Me) , − C H 2 C H 3 (− Et).) (n = 1–10). [ABSTRACT FROM AUTHOR]

We have investigated the basic mechanism of carbon nanotube (CNT) interactions with various room-temperature ionic liquids (RTILs) using molecular dynamics (MD) simulations. To understand the effects of the cation molecular geometry on the properties of the interface structure in the RTIL systems, we have studied a set of three RTILs with the same [BF4]- (tetrafluoroborate) anion but with different cations, namely, [EMIM]+ (1-ethyl-3-methylimidazolium), [BMIM]+ (1-butyl-3-methylimidazolium), [HMIM]+ (1-hexyl-3-methylimidazolium), and [OMIM]+ (1-octyl-3-methylimidazolium) ions. The simulation results showed that the imidazolium cations exhibit two distinct orientations (perpendicular and parallel to the CNTs surface) at the interface irrespective of the alkyl chain length of the cations. The average number of hydrogen bonds per cations inside the CNT was found to be higher for [OMIM][BF4] (1.01), which suggests that [OMIM]+ imidazolium rings to be concentrated at the center of the CNT, which favors hydrogen bond. The reported results show the diffusion coefficients of ions in confinement are much lower in comparison to the bulk region. The interaction energy between [OMIM][BF4] (-8.75 kcal.mol−1.ion−1) and CNT was found to be higher as compared to other ILs. The cations paralleling the CNT surface are thermodynamically significantly more stable because of the substantial interfacial π-π stacking interactions, as shown by a comparison with the calculated interaction energies between cations and the CNTs. Our simulation results provide a molecular-level understanding of the stabilization and dispersion of CNT bundles in ILs. [ABSTRACT FROM AUTHOR]

This study investigates the interaction between copper (Cu) and silver (Ag) clusters and graphene-based materials using molecular dynamics simulations. It focuses on how graphene oxidation and aminated polyethylene glycol (PEG-NH2) functionalization influence interaction strength and cluster dynamics. The analysis includes pristine graphene (PG), low-oxidized graphene oxide (GOL), and PEGylated graphene oxide (GO-PEG-NH2). The results reveal that clusters on PG exhibit high mobility, while GO-PEG-NH2 significantly restricts mobility due to strong interactions, as evidenced by highly negative interaction energies. GO-PEG-NH2 systems also display pronounced subdiffusive behavior (α < 1), indicating strong binding and constrained motion. These findings underscore the critical role of PEG-NH2 functionalization in controlling cluster diffusion, paving the way for innovative designs in biomedical and catalytic nanocarrier applications. [ABSTRACT FROM AUTHOR]

Starting from (p‐tolylsulfinyl)ferrocene (1), a mixture of the complete series [CpFe{C5H5–n(SOTol‐p)n}] (n = 2–4) (2–4) in all regioisomers was obtained. After chromatographic separation, crystals of 1,2‐bis[(4‐methylbenzene)sulfinyl]ferrocene, 2a, and 1,3‐bis[(4‐methylbenzene)sulfinyl]ferrocene, 2b, both [Fe(C5H5)(C19H17O2S2)], as well as of 1,2,3‐tris[(4‐methylbenzene)sulfinyl]ferrocene, [Fe(C5H5)(C26H23O3S3)], 3a, and 1,2,3,4‐tetrakis[(4‐methylbenzene)sulfinyl]ferrocene ethyl acetate 0.75‐solvate, [Fe(C5H5)(C33H29O4S4)]·0.75C4H8O2, 4, could be isolated. Their molecular and crystal structures are compared with each other and also with the so far unreported structures of related 1,2‐bis(phenylsulfanyl)ferrocene, [Fe(C5H5)(C17H13S2)], 5, and 1,2,3,4‐tetrakis(phenylsulfanyl)ferrocene, [Fe(C5H5)(C29H21S4)], 6. In all the sulfinyl structures, the O atoms of the S=O groups are in equatorial positions, except for that in tetrasubstituted 4. All the arene rings of these compounds (except for one ring in 4) are in axial positions directed away from the Fe atom, mostly in a near perpendicular orientation with respect to the plane of the cyclopentadienyl ring. The main intermolecular interactions in the crystals are C—H...H—C, C—H...π and C—H...O, while C—H...S interactions are much less important, except for tetrasulfanyl compound 6. π–π interactions (intramolecular) are only important in compound 3a. Hirshfeld analysis shows that dispersion terms are dominant for the interaction energies of all six compounds. In general, the calculated total interaction energies increase with increasing number of substituents and are higher for the sulfinyl than for the sulfanyl groups. [ABSTRACT FROM AUTHOR]

This work conducts a comprehensive theoretical study on the non‐covalent complexation between cyclocarbons and C60 fullerene for the first time. The binding energy between cyclocarbons and C60 fullerene is significantly stronger than that between two C18 or two C60 fullerenes, indicating a particularly strong affinity. The cyclocarbons and C60 fullerene can spontaneously assemble into complexes in the gas phase at room temperature, and the hydrophobic effect caused by the solvent environment can promote this binding. The binding strength with C60 fullerene increases almost linearly with the increase of cyclocarbon size, and the C34@C60 dimer exhibits a perfect nano‐Saturn structure. As the ring size increases, the angle between the two cyclocarbons of the 2 : 1 trimers formed by cyclocarbons and C60 fullerene gradually decreases. In C60@2 C34 trimer, the fullerene is symmetrically surrounded by two cyclocarbons. The results on the trimers formed by cyclocarbon and C60 fullerenes in a 1 : 2 ratio showed when the cyclocarbon sandwiched between two fullerenes is not quite large, the trimers exhibit an ideal dumbbell‐like structure, and the presence of the first fullerene has a significant synergistic effect on the binding of the second one. The cyclocarbon greatly promotes the dimerization of fullerenes, which acted as a "molecular glue". [ABSTRACT FROM AUTHOR]

To reveal the mechanism of interaction between styrene−butadiene rubber (SBR) modifier and asphalt components, the molecular dynamics (MD) method was used to study the diffusion behaviour at different temperatures. The diffusion behaviour was discussed using radial distribution function (RDF), mean square displacement (MSD), and relative concentration (RC). Interaction energy was used to verify the accuracy of the simulations. The simulated results indicate that the order of the four asphalt components prone to aggregation with SBR is as follows: saturated >aromatics >asphaltene >resins. The interaction between SBR modifier and asphalt components affects the diffusion of asphalt components. At 160°C, the distribution of components is the most uniform. When the temperature is too low or too high, the components dissociate from each other. The results of interaction energy are consistent with the conclusion. This study reveals the mechanism of component interaction of SBR modified asphalt. [ABSTRACT FROM AUTHOR]

In order to explore the micro-mechanism of the influence of anionic surfactant and nonionic surfactant on the wettability of low-rank coal, the molecular dynamics simulation method (MD) was used to study the influence of anionic surfactant (SDS), nonionic surfactant (AEO3) and their complex system on the wettability of lignite. The results show that when SDS is compounded with AEO3, the thickness of coal/water adsorption layer is 36.35×10−10 m, which wets 58.83% of lignite molecular layer. At this time, the diffusion coefficient of water molecules increases significantly, thus increasing the probability of collision between coal and water molecules, and further enhancing the wettability of water on lignite. In SDS/ AEO3 compound system, the interaction energy between lignite and water and between surfactant and water is greater than that of SDS and AEO3 alone. The interaction energy between coal and water, surfactant and water are all provided by non-bond energy, and the electrostatic energy accounts for more than 82.28% of the non-bond energy, while the contribution of Van der Waals’ force energy is very small, even inhibiting the interaction between surfactant and water. There is little difference in the length of hydrogen bonds between SDS, AEO3 and SDS/ AEO3, but the number of hydrogen bonds between SDS/ AEO3 and water is more than that SDS and AEO3 separated produced with water.

This is an article in the field of metallurgical engineering. In order to explore the leaching mechanism of zinc oxide from zinc-containing dust by choline chloride and oxalic acid, the interaction of choline chloride and oxalic acid with 1∶1 and 1∶2 eutectic solvent on zinc oxide surface was simulated by means of quantum mechanics, and the simulation results were verified by experiments. The results show that the intermolecular hydrogen bond formed by choline chloride and oxalic acid plays an important role in the stability of eutectic solvent. Ch+ gains electrons, OA and Cl- lose electrons. ZnO(001) surface is the complete cleavage surface of zinc oxide, oxygen and zinc atoms form tetrahedral structure, oxygen atom is easy to gain electrons and zinc atom is easy to lose electrons during the reaction with the outside world. The zinc oxide loses electrons during the interaction between cholinergic chloride-oxalic acid and zinc oxide, and cholinergic chloride-oxalic acid gains electrons. The interaction energy △E=-819.6896Ha between ChCl-2OA and ZnO is small, indicating that zinc oxide is more likely to react with ChCl-2OA. The radial distribution function shows that the chemical adsorption is the main factor in the interaction between choline chloride and oxalic acid and zinc oxide, and the physical adsorption is the secondary factor. The contribution of Cl-Zn formed by chemical adsorption is greater than that of O-Zn. Through the pure mineral test, it can be verified that ChCl-2OA has a better leaching effect on zinc oxide, which verifies the accuracy of molecular simulation of the interaction mechanism between choline chloride and oxalic acid and zinc oxide, and provides a theoretical basis for the leaching of zinc-containing dust with eutectic solvent.

Solid-state drug delivery systems for the drug substances transport are of great importance nowadays. In the present work, the non-covalent interactions between taxifolin (Tax) and graphene as well as nitrogenated (N-doped) graphenes were systematically studied by using a wide set of theoretical techniques. Symmetry-adapted perturbation theory (SAPT0) calculations confirmed more favorable adsorption of Tax on N-doped graphenes compared to pristine graphene. It was established that dispersion interactions play the main role in the attractive interactions (>60%), whereas electrostatic and induction forces contribute only moderately to the attraction (~25% and 7–8%, respectively). Independent gradient model (IGM) analysis visually demonstrated the existence of dispersion interactions and hydrogen bonding in the studied Tax complexes. Ab initio molecular dynamics calculations indicated stability of these complexes at different temperatures. Our results show that N-doped graphenes with the enhanced interaction energy (Eint) toward Tax are promising candidates for the technical realization of the targeted drug delivery systems.

The heat of mixing of some petrological relevant substitutions (i.e., Mg-Al, Si-Al, Mg-Ti, Mg-Ca, and Mg-Fe) was investigated systematically in silicates, titanates, tungstates, carbonates, oxides, hydroxides, and sulphates by density functional theory calculations (e.g., melilite, chlorite, biotite, brucite, cordierite, amphibole, talc, pseudobrookite, pyroxene, olivine, wadsleyite, ilmenite, MgWO4, ringwoodite (spinel), perovskite, pyrope-grossular, magnesite-calcite, MgO-CaO, anhydrous and different hydrated MgSO4). A specific substitution is characterised by different microscopic interaction energies in different minerals, e.g., the octahedral Mg-Al exchange on a single crystallographic site in pyroxene has a microscopic interaction energy that is more than twice compared to that in biotite. A comparative investigation of the heat of mixing using microscopic interaction energies on a single crystallographic site has the advantage that they are not influenced by cation ordering. They could be successfully correlated with the stiffnesses of the minerals, which in turn were scaled to the oxygen packing fraction, a parameter that is easily available for poorly investigated minerals. With this information, the interaction energies of a certain substitution can be transferred from minerals where they are well-known to mineral groups where they are less- or unknown. Using the cross-site terms and the microscopic interaction energies, the macroscopic interaction energies of the coupled substitution, e.g., Mg + Si = Al + Al, of biotite and pyroxene were calculated, which are, however, affected by cation ordering and different degrees of local charge balance, for which appropriate models are necessary. [ABSTRACT FROM AUTHOR]

Asphaltene deposition poses a significant threat to the safe production of crude oil as it tends to occur in forming porous media, wellbores, pipelines, and refining equipment. It is also responsible for the high viscosity of crude oil, making it challenging to recover heavy crude oil. This study investigates the impact of aliphatic chain length on asphaltene aggregation and the molecular properties of organic solvents and dispersants interacting with asphaltenes. To achieve this, two model asphaltenes, M1 and C5Pe, were used in molecular dynamics simulations using radial distribution function, root means square displacement, diffusion coefficient, solubility parameter, cohesion energy density, and interaction energy as characterization parameters. The simulations revealed that the aggregation onset of C5Pe and M1 occurs at 2–5 Å at room temperature and pressure, and π–π stacking is the primary mode of aggregation, with the presence of heteroatomic groups providing additional forces for stable aggregation. Additionally, it was found that the π–π interactions weakened with the increase of the side chain length. The study also investigated the molecular behavior of asphaltenes in six organic solvents and found that M1 was more soluble in aromatic solvents than C5Pe, and undesirable solvents could induce asphaltene aggregation. Furthermore, the effect of five dispersants on asphaltene aggregation was explored, with PIBSI, BEHP, B-DBSA, CTAB, and benzoic acid showing decreasing dispersing ability in that order at lower concentrations. However, at higher concentrations, the effect of some dispersants diminishes and triggers self-aggregation behavior. [ABSTRACT FROM AUTHOR]

To solve the problem of the oscillation stability of direct-driven wind-powered grid-connected systems with a low-voltage crossing control, a method of oscillation stability analysis for these systems, based on interactive energy, is proposed in this paper. Firstly, a dynamic energy model of a direct-driven wind-powered grid-connected system is established, considering a low-voltage traverse control. Secondly, the energy is divided into four parts—the line parameter energy, current inner loop energy, PLL energy, and current inner loop–PLL interaction energy—and the conduction path of the energy during a low-voltage crossing is described. On this basis, the aperiodic components of each energy path are analyzed, the stability level of the system is quantified, the influence of the different control parameters on the interactive energy dissipation is deduced, the key interactive control links affecting the stability of the system are screened, and the influence rules of the parameters are expounded. Finally, a direct-driven wind-powered grid-connected system model is built on the Rt-lab platform, and it is verified by a simulation test. The results show that the interaction energy generated by the interaction of the current inner loop and phase-locked loop is a key factor affecting the stability of the direct-driven wind-powered grid-connected system. The simulation test parameters of the control group were adjusted as the current inner loop's proportion parameter increased from 1.32 to 5.28, the current inner loop's integral parameter increased from 4.48 to 6.42, the PLL's proportion parameter decreased from 9.45 to 6.3, and the PLL's integral parameter decreased from 50.25 to 40.2. Both the theoretical and experimental results show that increasing the current inner loop's integral and proportion parameters can improve the stability level of the direct-driven wind-powered grid-connected system; reducing the phase-locked loop's proportion and integral parameters can also improve the stability level of the grid-connected system. [ABSTRACT FROM AUTHOR]

Solid-state drug delivery systems for the drug substances transport are of great importance nowadays. In the present work, the non-covalent interactions between taxifolin (Tax) and graphene as well as nitrogenated (N-doped) graphenes were systematically studied by using a wide set of theoretical techniques. Symmetry-adapted perturbation theory (SAPT0) calculations confirmed more favorable adsorption of Tax on N-doped graphenes compared to pristine graphene. It was established that dispersion interactions play the main role in the attractive interactions (>60%), whereas electrostatic and induction forces contribute only moderately to the attraction (~25% and 7–8%, respectively). Independent gradient model (IGM) analysis visually demonstrated the existence of dispersion interactions and hydrogen bonding in the studied Tax complexes. Ab initio molecular dynamics calculations indicated stability of these complexes at different temperatures. Our results show that N-doped graphenes with the enhanced interaction energy (Eint) toward Tax are promising candidates for the technical realization of the targeted drug delivery systems. [ABSTRACT FROM AUTHOR]

Determining binding affinities in protein–protein and protein–peptide complexes is a challenging task that directly impacts the development of peptide and protein pharmaceuticals. Although several models have been proposed to predict the value of the dissociation constant and the Gibbs free energy, they are currently not capable of making stable predictions with high accuracy, in particular for complexes consisting of more than two molecules. In this work, we present ProBAN, a new method for predicting binding affinity in protein–protein complexes based on a deep convolutional neural network. Prediction is carried out for the spatial structures of complexes, presented in the format of a 4D tensor, which includes information about the location of atoms and their abilities to participate in various types of interactions realized in protein–protein and protein–peptide complexes. The effectiveness of the model was assessed both on an internal test data set containing complexes consisting of three or more molecules, as well as on an external test for the PPI‐Affinity service. As a result, we managed to achieve the best prediction quality on these data sets among all the analyzed models: on the internal test, Pearson correlation R = 0.6, MAE = 1.60, on the external test, R = 0.55, MAE = 1.75. The open‐source code, the trained ProBAN model, and the collected dataset are freely available at the following link https://github.com/EABogdanova/ProBAN. [ABSTRACT FROM AUTHOR]