Search

Determining the catalytic site of enzymes is a great help for understanding the relationship between protein sequence, structure, and function, which provides the basis and targets for designing, modifying, and enhancing enzyme activity. The unique local spatial configuration bound to the substrate at the active center of the enzyme determines the catalytic ability of enzymes and plays an important role in the catalytic site prediction. As a suitable tool, the graph neural network can better understand and identify the residue sites with unique local spatial configurations due to its remarkable ability to characterize the three-dimensional structural features of proteins. Consequently, a novel model for predicting enzyme catalytic sites has been developed, which incorporates a uniquely designed adaptive edge-gated graph attention neural network (AEGAN). This model is capable of effectively handling sequential and structural characteristics of proteins at various levels, and the extracted features enable an accurate description of the local spatial configuration of the enzyme active site by sampling the local space around candidate residues and special design of amino acid physical and chemical properties. To evaluate its performance, the model was compared with existing catalytic site prediction models using different benchmark datasets and achieved the best results on each benchmark dataset. The model exhibited a sensitivity of 0.9659, accuracy of 0.9226, and area under the precision-recall curve (AUPRC) of 0.9241 on the independent test set constructed for evaluation. Furthermore, the F1-score of this model is nearly four times higher than that of the best-performing similar model in previous studies. This research can serve as a valuable tool to help researchers understand protein sequence–structure–function relationships while facilitating the characterization of novel enzymes of unknown function.

A feasible and promising carbon reduction strategy is to convert CO2into C1 compounds through renewable energy-driven first and then use a multi-enzyme cascade to obtain compounds with higher carbon numbers and market value. In this study, three multi-enzyme cascade pathways for preparing important C2/C4 compounds (ethylene glycol [EG], glycolic acid [GA], and d-erythrose) from methanol were proposed. All pathways use glycolaldehyde as the key intermediate, which can be obtained from methanol through glycolaldehyde synthase and alcohol oxidase cascade. Then, the target compounds can be acquired by adding different enzymes. As a result, 0.90 g/L EG, 0.33 g/L GA, and 53.65 mg/L d-erythrose were produced from methanol. Compared with the titers of EG (6.6 mM) and GA (1.2 g/L) synthesis routes from formaldehyde reported in the literature, the titers of this study (27.58 mM EG and 1.59 g/L GA from formaldehyde) were significantly improved.

The core issue in implementing in silicoenzyme screening lies in accurately evaluating the merits of mutants. The best solution to this problem would undoubtedly be the precise prediction of kinetic parameters for mutant enzymes to directly assess the catalytic efficiency and activity of enzymes. Previously developed models of this type are mostly limited to predictions for wild-type enzymes and tend to exhibit poorer generalization capabilities. Here, a novel deep-learning model framework and an ensemble iterative transfer learning strategy for enzyme mutant kinetics parameter (kcat, Km, and KKm) prediction (EITLEM-Kinetics) were developed. This approach is designed to overcome the limitations imposed by sparse training samples on the model’s predictive performance and accurately predict the kinetic parameters of various mutants. This development is set to provide significant assistance in future endeavors to construct virtual screening methods aimed at enhancing enzyme activity and offer innovative solutions for researchers grappling with similar challenges.

Adipic acid has extensive industrial applications in various fields. The direct production of adipic acid by the bioreduction of glucose-derived cis,cis-muconic acid constitutes a remarkably promising pathway. While the enoate reductase from Heyndrickxia coagulanspossesses weak catalytic activity in the conversion of cis,cis-muconic acid to adipic acid, its low activity makes it challenging to meet the demands of industrial production. In this study, we devised an effective multiscale computational strategy to design variants with enhanced activity. A focused library comprising only 15 variants was generated. Experimental validation confirmed that the in vivoactivity of the optimal variant M8(G27M/I374S) was enhanced by 14.2-fold compared to the wild-type enzyme. Adipic acid production reached 0.9 g/L under 10 mM substrate concentration, achieving a yield of 61%, which is about 19 times higher than the wild type (3.2%). Further engineering of metabolic pathways offers promising prospects for achieving industrial biosynthesis of adipic acid.

High energy density fuels (HEDFs) that have high volumetric net heat of combustion (NHOC), high stability, and high environmental resistance are greatly important in the fuel field and in military bases and aerospace applications. In this paper, molecular dynamics and quantum chemistry were used to compute the significant physical properties of candidate molecules for HEDFs, such as their enthalpies of combustion, enthalpies of vaporization, densities, and melting points. A computational protocol for evaluating these properties in the fuel field was established, including a new method for estimating the melting point. By using our protocol, we found that, to improve fuel performances such as the density and volumetric NHOC, cyclopropanation is obviously better than hydrogenation. Our protocol was verified to have good accuracy and can be used to compare and assess different target molecules as potential HEDFs.

Charge transfer (CT) was proposed to play a role in the cyclodextrin (CD) complexation with guest molecules. To elucidate the importance of CT interactions, here we used computational methods of quantum mechanics, docking, and molecular dynamics (MD) to investigate α-CD complexes with aromatic guest molecules of nitrobenzene, carboxybenzene, benzoate, 4-nitrophenol, and 4-nitrophenolate. Considering host–guest CT in the docking has more of a chance to search reasonable guest orientations relative to α-CD matching the experiment, compared to that without CT. The CT interaction enlarges the difference in binding affinities of varied guests, as evidenced from potential of mean force (PMF) MD calculations. Energy decomposition of the total enthalpy and entropy shows the CT influence on the binding reactions in detail and indicates that there are considerable compensating effects of individual contributions from the binding partners and surrounding water. Charge transfer reduces the total dipole of α-CD by 9% on average and alters its dipole direction thereby affecting guest association. Gas-phase zeroth-order symmetry-adapted perturbation theory calculations show host–guest CT amounts to approximately 6% of the total binding energy. The continuum solvation model based on the quantum mechanical charge density predicts binding energies comparable with the well depth of PMF profiles in explicit water. The abnormal binding strength of α-CD with the similar guests can be rationalized in terms of hydrogen bonding, extent of host–guest CT, and dipole arrangement of guest relative to host.

For the purpose of industrial applications, deactivation behavior of the hydroxyapatite catalyst was studied in the dehydration of lactic acid to acrylic acid. The hydroxyapatite catalyst shows stable acrylic acid selectivity with 140 h on stream. Obvious increases in acetaldehyde and lactide selectivities were observed from 140 to 250 h, which resulted in a decrease of sharp acrylic acid selectivity. The fresh and used hydroxyapatite catalysts were further characterized by techniques including N2adsorption–desorption, X-ray diffraction, transmission electron microscopy, 13C cross-polarization magic-angle spinning nuclear magnetic resonance, thermal gravity analysis, and temperature-programmed desorption of CO2and NH3. These results indicate that the mechanism of coke deactivation consists of two steps. One is the formation of aliphatic carbon by the deposition of low carbon products with 140 h; the other is the formation of aromatic coke by the condensation of lactide. Also, the best generated time of 140 h was determined.

The partitioning of amino acids between water and apolar environments is of vital importance in protein function and drug delivery. Here we present an extensive benchmark for octanol/water (log Poct), chloroform/water (log Pclf), and cyclohexane/water (log Pchx) partition coefficients of neutral amino acid side chain analogues (SCAs) with Amber families of ff99SB-ILDN, ff03, ff14SB, fb15, and ff15ipq, CHARMM 27, GROMOS 53A6, and OPLS-AA/L force fields. A root-mean-square error (RMSE) of 0.4–1.3 log units from experiment is observed for the tested FFs, of which Amber ff94 lineages of ff99SB-ILDN, ff14SB, and fb15 perform best with an RMSE and mean signed error (MSE) of about 0.5 and 0.2 log units, respectively, a performance comparable with quantum mechanical SMD calculations. This finding retains the possibility of modeling proteins in varied environments with one set of classical molecular mechanical force fields. All the FFs tend to overestimate log P, except for GROMOS 53A6 underestimating log Pclfand log Pchx. These discrepancies are mainly due to the larger overestimated solvation free energies in water (ΔGwat) relative to that in organic solvents (ΔGoct, ΔGclf, and ΔGchx); for GROMOS 53A6, it is due to the underestimated ΔGwatand ΔGoct. The latest water models of “FB” and “OPC” families paired with the recent Amber fb15 do not show an obvious improvement for ΔGwatand log Pcalculations. The van der Waals interaction between amino acids and cyclohexane is found to be too strong (overestimated) systematically. Scaling protein–water interactions lead to more favorable ΔGwat, thereby lowering log Pand resulting in a better performance for Amber ff03ws, while such scaling seems a bit too much for Amber ff99SBws. This, along with our previous work (Zhang et al. J. Chem. Inf. Model.2018, 58, 1037−1052), may aid in the development and systematic improvements of classical force fields to model proteins in aqueous and nonaqueous phases accurately.

In memory-based collaborative filtering (CF) algorithms, the similarity and prediction method have a significant impact on the recommendation results. Most of the existing recommendation techniques have improved different similarity measures to alleviate inaccurate similarity results in sparse data, however, ignored the impact of sparse data on prediction results. To enhance the adaptability to sparse data, we propose a new item-based CF algorithm, which consists of the item similarity measure based vague sets and item-based prediction method with the new neighbor selection strategy. First, in the stage of similarity calculation, the Kullback–Leibler (KL) divergence based on vague sets is proposed from the perspective of user preference probability to measure item similarity. Following this, the impact of rating quantity is further considered to improve the accuracy of similarity results. Next, in the prediction stage, we relax the limit of depending on explicitly ratings and integrate more rating information to adjust prediction results. Experimental results on benchmark data sets show that, compared with other representative algorithms, our algorithm has better prediction and recommendation quality, and effectively alleviates the data sparseness problem.

Biomanufacturing, which uses renewable resources as raw materials and uses biological processes to produce energy and chemicals, has long been regarded as a production model that replaces the unsustainable fossil economy. The construction of non-natural and efficient biosynthesis routes of chemicals is an important goal of green biomanufacturing. Traditional methods that rely on experience are difficult to support the realization of this goal. However, with the rapid development of information technology, the intelligence of biomanufacturing has brought hope to achieve this goal. Retrobiosynthesis and computational enzyme design, as two of the main technologies in intelligent biomanufacturing, have developed rapidly in recent years and have made great achievements and some representative works have demonstrated the great value that the integration of the two fields may bring. To achieve the final integration of the two fields, it is necessary to examine the information, methods and tools from a bird's-eye view, and to find a feasible idea and solution for establishing a connection point. For this purpose, this article briefly reviewed the main ideas, methods and tools of the two fields, and put forward views on how to achieve the integration of the two fields.