Prediction of protein-ligand binding affinity via deep learning models.

Accurately predicting the binding affinity between proteins and ligands is crucial in drug screening and optimization, but it is still a challenge in computer-aided drug design. The recent success of AlphaFold2 in predicting protein structures has brought new hope for deep learning (DL) models to accurately predict protein-ligand binding affinity. However, the current DL models still face limitations due to the low-quality database, inaccurate input representation and inappropriate model architecture. In this work, we review the computational methods, specifically DL-based models, used to predict protein-ligand binding affinity. We start with a brief introduction to protein-ligand binding affinity and the traditional computational methods used to calculate them. We then introduce the basic principles of DL models for predicting protein-ligand binding affinity. Next, we review the commonly used databases, input representations and DL models in this field. Finally, we discuss the potential challenges and future work in accurately predicting protein-ligand binding affinity via DL models.

Multi-task bioassay pre-training for protein-ligand binding affinity prediction.

Protein-ligand binding affinity (PLBA) prediction is the fundamental task in drug discovery. Recently, various deep learning-based models predict binding affinity by incorporating the three-dimensional (3D) structure of protein-ligand complexes as input and achieving astounding progress. However, due to the scarcity of high-quality training data, the generalization ability of current models is still limited. Although there is a vast amount of affinity data available in large-scale databases such as ChEMBL, issues such as inconsistent affinity measurement labels (i.e. IC50, Ki, Kd), different experimental conditions, and the lack of available 3D binding structures complicate the development of high-precision affinity prediction models using these data. To address these issues, we (i) propose Multi-task Bioassay Pre-training (MBP), a pre-training framework for structure-based PLBA prediction; (ii) construct a pre-training dataset called ChEMBL-Dock with more than 300k experimentally measured affinity labels and about 2.8M docked 3D structures. By introducing multi-task pre-training to treat the prediction of different affinity labels as different tasks and classifying relative rankings between samples from the same bioassay, MBP learns robust and transferrable structural knowledge from our new ChEMBL-Dock dataset with varied and noisy labels. Experiments substantiate the capability of MBP on the structure-based PLBA prediction task. To the best of our knowledge, MBP is the first affinity pre-training model and shows great potential for future development. MBP web-server is now available for free at: https://huggingface.co/spaces/jiaxianustc/mbp.

Thermodynamic consequences of stapling side-chains on a peptide ligand using a lactam-bridge: A theoretical study on anti-angiogenic peptides targeting VEGF.

Peptides are promising therapeutic agents for various biological targets due to their high efficacy and low toxicity, and the design of peptide ligands with high binding affinity to the target of interest is of utmost importance in peptide-based drug design. Introducing a conformational constraint to a flexible peptide ligand using a side-chain lactam-bridge is a convenient and efficient method to improve its binding affinity to the target. However, in general, such a small structural modification to a flexible ligand made with the intent of lowering the configurational entropic penalty for binding may have unintended consequences in different components of the binding enthalpy and entropy, including the configurational entropy component, which are still not clearly understood. Toward probing this, we examine different components of the binding enthalpy and entropy as well as the underlying structure and dynamics, for a side-chain lactam-bridged peptide inhibitor and its flexible analog forming complexes with vascular endothelial growth factor (VEGF), using all-atom molecular dynamics simulations. It is found that introducing a side-chain lactam-bridge constraint into the flexible peptide analog led to a gain in configurational entropy change but losses in solvation entropy, solute internal energy, and solvation energy changes upon binding, pinpointing the opportunities and challenges in drug design. The present study features an interplay between configurational and solvation entropy changes, as well as the one between binding enthalpy and entropy, in ligand-target binding upon imposing a conformational constraint into a flexible ligand.

Molecular persistent spectral image (Mol-PSI) representation for machine learning models in drug design.

Artificial intelligence (AI)-based drug design has great promise to fundamentally change the landscape of the pharmaceutical industry. Even though there are great progress from handcrafted feature-based machine learning models, 3D convolutional neural networks (CNNs) and graph neural networks, effective and efficient representations that characterize the structural, physical, chemical and biological properties of molecular structures and interactions remain to be a great challenge. Here, we propose an equal-sized molecular 2D image representation, known as the molecular persistent spectral image (Mol-PSI), and combine it with CNN model for AI-based drug design. Mol-PSI provides a unique one-to-one image representation for molecular structures and interactions. In general, deep models are empowered to achieve better performance with systematically organized representations in image format. A well-designed parallel CNN architecture for adapting Mol-PSIs is developed for protein-ligand binding affinity prediction. Our results, for the three most commonly used databases, including PDBbind-v2007, PDBbind-v2013 and PDBbind-v2016, are better than all traditional machine learning models, as far as we know. Our Mol-PSI model provides a powerful molecular representation that can be widely used in AI-based drug design and molecular data analysis.

fastDRH: a webserver to predict and analyze protein-ligand complexes based on molecular docking and MM/PB(GB)SA computation.

Predicting the native or near-native binding pose of a small molecule within a protein binding pocket is an extremely important task in structure-based drug design, especially in the hit-to-lead and lead optimization phases. In this study, fastDRH, a free and open accessed web server, was developed to predict and analyze protein-ligand complex structures. In fastDRH server, AutoDock Vina and AutoDock-GPU docking engines, structure-truncated MM/PB(GB)SA free energy calculation procedures and multiple poses based per-residue energy decomposition analysis were well integrated into a user-friendly and multifunctional online platform. Benefit from the modular architecture, users can flexibly use one or more of three features, including molecular docking, docking pose rescoring and hotspot residue prediction, to obtain the key information clearly based on a result analysis panel supported by 3Dmol.js and Apache ECharts. In terms of protein-ligand binding mode prediction, the integrated structure-truncated MM/PB(GB)SA rescoring procedures exhibit a success rate of >80% in benchmark, which is much better than the AutoDock Vina (~70%). For hotspot residue identification, our multiple poses based per-residue energy decomposition analysis strategy is a more reliable solution than the one using only a single pose, and the performance of our solution has been experimentally validated in several drug discovery projects. To summarize, the fastDRH server is a useful tool for predicting the ligand binding mode and the hotspot residue of protein for ligand binding. The fastDRH server is accessible free of charge at http://cadd.zju.edu.cn/fastdrh/.

Multiphysical graph neural network (MP-GNN) for COVID-19 drug design.

Graph neural networks (GNNs) are the most promising deep learning models that can revolutionize non-Euclidean data analysis. However, their full potential is severely curtailed by poorly represented molecular graphs and features. Here, we propose a multiphysical graph neural network (MP-GNN) model based on the developed multiphysical molecular graph representation and featurization. All kinds of molecular interactions, between different atom types and at different scales, are systematically represented by a series of scale-specific and element-specific graphs with distance-related node features. From these graphs, graph convolution network (GCN) models are constructed with specially designed weight-sharing architectures. Base learners are constructed from GCN models from different elements at different scales, and further consolidated together using both one-scale and multi-scale ensemble learning schemes. Our MP-GNN has two distinct properties. First, our MP-GNN incorporates multiscale interactions using more than one molecular graph. Atomic interactions from various different scales are not modeled by one specific graph (as in traditional GNNs), instead they are represented by a series of graphs at different scales. Second, it is free from the complicated feature generation process as in conventional GNN methods. In our MP-GNN, various atom interactions are embedded into element-specific graph representations with only distance-related node features. A unique GNN architecture is designed to incorporate all the information into a consolidated model. Our MP-GNN has been extensively validated on the widely used benchmark test datasets from PDBbind, including PDBbind-v2007, PDBbind-v2013 and PDBbind-v2016. Our model can outperform all existing models as far as we know. Further, our MP-GNN is used in coronavirus disease 2019 drug design. Based on a dataset with 185 complexes of inhibitors for severe acute respiratory syndrome coronavirus (SARS-CoV/SARS-CoV-2), we evaluate their binding affinities using our MP-GNN. It has been found that our MP-GNN is of high accuracy. This demonstrates the great potential of our MP-GNN for the screening of potential drugs for SARS-CoV-2. Availability: The Multiphysical graph neural network (MP-GNN) model can be found in https://github.com/Alibaba-DAMO-DrugAI/MGNN. Additional data or code will be available upon reasonable request.

Insights into Modeling Approaches in Chemistry: Assessing Ligand-Protein Binding Thermodynamics Based on Rigid-Flexible Model Molecules.

In the field of chemistry, model compounds find extensive use for investigating complex objects. One prime example of such object is the protein-ligand supramolecular interaction. Prediction the enthalpic and entropic contribution to the free energy associated with this process, as well as the structural and dynamic characteristics of protein-ligand complexes poses considerable challenges. This review exemplifies modeling approaches used to study protein-ligand binding (PLB) thermodynamics by employing pairs of conformationally constrained/flexible model molecules. Strategically designing the model molecules can reduce the number of variables that influence thermodynamic parameters. This enables scientists to gain deeper insights into the enthalpy and entropy of PLB, which is relevant for medicinal chemistry and drug design. The model studies reviewed here demonstrate that rigidifying ligands may induce compensating changes in the enthalpy and entropy of binding. Some "rules of thumb" have started to emerge on how to minimize entropy-enthalpy compensation and design efficient rigidified or flexible ligands.

The Three-Dimensional Reference Interaction Site Model Approach as a Promising Tool for Studying Hydrated Viruses and Their Complexes with Ligands.

Viruses are the most numerous biological form living in any ecosystem. Viral diseases affect not only people but also representatives of fauna and flora. The latest pandemic has shown how important it is for the scientific community to respond quickly to the challenge, including critically assessing the viral threat and developing appropriate measures to counter this threat. Scientists around the world are making enormous efforts to solve these problems. In silico methods, which allow quite rapid obtention of, in many cases, accurate information in this field, are effective tools for the description of various aspects of virus activity, including virus-host cell interactions, and, thus, can provide a molecular insight into the mechanism of virus functioning. The three-dimensional reference interaction site model (3D-RISM) seems to be one of the most effective and inexpensive methods to compute hydrated viruses, since the method allows us to provide efficient calculations of hydrated viruses, remaining all molecular details of the liquid environment and virus structure. The pandemic challenge has resulted in a fast increase in the number of 3D-RISM calculations devoted to hydrated viruses. To provide readers with a summary of this literature, we present a systematic overview of the 3D-RISM calculations, covering the period since 2010. We discuss various biophysical aspects of the 3D-RISM results and demonstrate capabilities, limitations, achievements, and prospects of the method using examples of viruses such as influenza, hepatitis, and SARS-CoV-2 viruses.

Rapid Determination of Kinetic Constants for Slow-Binding Inhibitors and Inactivators of Human Histone Deacetylase 8.

The kinetics and mechanism of drug binding to its target are critical to pharmacological efficacy. A high throughput (HTS) screen often results in hundreds of hits, of which usually only simple IC50 values are determined during reconfirmation. However, kinetic parameters such as residence time for reversible inhibitors and the kinact/KI ratio, which is the critical measure for evaluating covalent inactivators, are early predictive measures to assess the chances of success of the hits in the clinic. Using the promising cancer target human histone deacetylase 8 as an example, we present a robust method that calculates concentration-dependent apparent rate constants for the inhibition or inactivation of HDAC8 from dose-response curves recorded after different pre-incubation times. With these data, hit compounds can be classified according to their mechanism of action, and the relevant kinetic parameters can be calculated in a highly parallel fashion. HDAC8 inhibitors with known modes of action were correctly assigned to their mechanism, and the binding mechanisms of some hits from an internal HDAC8 screening campaign were newly determined. The oxonitriles SVE04 and SVE27 were classified as fast reversible HDAC8 inhibitors with moderate time-constant IC50 values of 4.2 and 2.6 µM, respectively. The hit compound TJ-19-24 and SAH03 behave like slow two-step inactivators or reversible inhibitors, with a very low reverse isomerization rate.

A Point Cloud Graph Neural Network for Protein-Ligand Binding Site Prediction.

Predicting protein-ligand binding sites is an integral part of structural biology and drug design. A comprehensive understanding of these binding sites is essential for advancing drug innovation, elucidating mechanisms of biological function, and exploring the nature of disease. However, accurately identifying protein-ligand binding sites remains a challenging task. To address this, we propose PGpocket, a geometric deep learning-based framework to improve protein-ligand binding site prediction. Initially, the protein surface is converted into a point cloud, and then the geometric and chemical properties of each point are calculated. Subsequently, the point cloud graph is constructed based on the inter-point distances, and the point cloud graph neural network (GNN) is applied to extract and analyze the protein surface information to predict potential binding sites. PGpocket is trained on the scPDB dataset, and its performance is verified on two independent test sets, Coach420 and HOLO4K. The results show that PGpocket achieves a 58% success rate on the Coach420 dataset and a 56% success rate on the HOLO4K dataset. These results surpass competing algorithms, demonstrating PGpocket's advancement and practicality for protein-ligand binding site prediction.

Protein-Ligand Binding and Structural Modelling Studies of Pheromone-Binding Protein-like Sol g 2.1 from Solenopsis geminata Fire Ant Venom.

Sol g 2 is the major protein in Solenopsis geminata fire ant venom. It shares the highest sequence identity with Sol i 2 (S. invicta) and shares high structural homology with LmaPBP (pheromone-binding protein (PBP) from the cockroach Leucophaea maderae). We examined the specific Sol g 2 protein ligands from fire ant venom. The results revealed that the protein naturally formed complexes with hydrocarbons, including decane, undecane, dodecane, and tridecane, in aqueous venom solutions. Decane showed the highest affinity binding (Kd) with the recombinant Sol g 2.1 protein (rSol g 2.1). Surprisingly, the mixture of alkanes exhibited a higher binding affinity with the rSol g 2.1 protein compared to a single one, which is related to molecular docking simulations, revealing allosteric binding sites in the Sol g 2.1 protein model. In the trail-following bioassay, we observed that a mixture of the protein sol g 2.1 and hydrocarbons elicited S. geminata worker ants to follow trails for a longer time and distance compared to a mixture containing only hydrocarbons. This suggests that Sol g 2.1 protein may delay the evaporation of the hydrocarbons. Interestingly, the piperidine alkaloids extracted have the highest attraction to the ants. Therefore, the mixture of hydrocarbons and piperidines had a synergistic effect on the trail-following of ants when both were added to the protein.

Molecular mechanism of binding between a therapeutic RNA aptamer and its protein target VEGF: A molecular dynamics study.

Macugen is a therapeutic RNA aptamer against vascular endothelial growth factor (VEGF)-165, the VEGF isoform primarily responsible for angiogenesis. It has been reported that Macugen inhibits angiogenesis by specifically binding to the heparin binding domain (HBD) of VEGF165. The mechanism of the molecular recognition between HBD and Macugen is investigated here using all-atom molecular dynamics simulations. We find that Macugen recognizes HBD by an induced-fit mechanism with major conformational changes in Macugen and almost no changes in the structure of HBD, whereas HBD recognizes Macugen by a conformational selection mechanism.

Computationally predicting binding affinity in protein-ligand complexes: free energy-based simulations and machine learning-based scoring functions.

Accurately predicting protein-ligand binding affinities can substantially facilitate the drug discovery process, but it remains as a difficult problem. To tackle the challenge, many computational methods have been proposed. Among these methods, free energy-based simulations and machine learning-based scoring functions can potentially provide accurate predictions. In this paper, we review these two classes of methods, following a number of thermodynamic cycles for the free energy-based simulations and a feature-representation taxonomy for the machine learning-based scoring functions. More recent deep learning-based predictions, where a hierarchy of feature representations are generally extracted, are also reviewed. Strengths and weaknesses of the two classes of methods, coupled with future directions for improvements, are comparatively discussed.

DeepDTAF: a deep learning method to predict protein-ligand binding affinity.

Biomolecular recognition between ligand and protein plays an essential role in drug discovery and development. However, it is extremely time and resource consuming to determine the protein-ligand binding affinity by experiments. At present, many computational methods have been proposed to predict binding affinity, most of which usually require protein 3D structures that are not often available. Therefore, new methods that can fully take advantage of sequence-level features are greatly needed to predict protein-ligand binding affinity and accelerate the drug discovery process. We developed a novel deep learning approach, named DeepDTAF, to predict the protein-ligand binding affinity. DeepDTAF was constructed by integrating local and global contextual features. More specifically, the protein-binding pocket, which possesses some special properties for directly binding the ligand, was firstly used as the local input feature for protein-ligand binding affinity prediction. Furthermore, dilated convolution was used to capture multiscale long-range interactions. We compared DeepDTAF with the recent state-of-art methods and analyzed the effectiveness of different parts of our model, the significant accuracy improvement showed that DeepDTAF was a reliable tool for affinity prediction. The resource codes and data are available at https: //github.com/KailiWang1/DeepDTAF.

Addressing data imbalance problems in ligand-binding site prediction using a variational autoencoder and a convolutional neural network.

Since 2015, a fast growing number of deep learning-based methods have been proposed for protein-ligand binding site prediction and many have achieved promising performance. These methods, however, neglect the imbalanced nature of binding site prediction problems. Traditional data-based approaches for handling data imbalance employ linear interpolation of minority class samples. Such approaches may not be fully exploited by deep neural networks on downstream tasks. We present a novel technique for balancing input classes by developing a deep neural network-based variational autoencoder (VAE) that aims to learn important attributes of the minority classes concerning nonlinear combinations. After learning, the trained VAE was used to generate new minority class samples that were later added to the original data to create a balanced dataset. Finally, a convolutional neural network was used for classification, for which we assumed that the nonlinearity could be fully integrated. As a case study, we applied our method to the identification of FAD- and FMN-binding sites of electron transport proteins. Compared with the best classifiers that use traditional machine learning algorithms, our models obtained a great improvement on sensitivity while maintaining similar or higher levels of accuracy and specificity. We also demonstrate that our method is better than other data imbalance handling techniques, such as SMOTE, ADASYN, and class weight adjustment. Additionally, our models also outperform existing predictors in predicting the same binding types. Our method is general and can be applied to other data types for prediction problems with moderate-to-heavy data imbalances.

Cheminformatics and Machine Learning Approaches to Assess Aquatic Toxicity Profiles of Fullerene Derivatives.

Fullerene derivatives (FDs) are widely used in nanomaterials production, the pharmaceutical industry and biomedicine. In the present study, we focused on the potential toxic effects of FDs on the aquatic environment. First, we analyzed the binding affinity of 169 FDs to 10 human proteins (1D6U, 1E3K, 1GOS, 1GS4, 1H82, 1OG5, 1UOM, 2F9Q, 2J0D, 3ERT) obtained from the Protein Data Bank (PDB) and showing high similarity to proteins from aquatic species. Then, the binding activity of 169 FDs to the enzyme acetylcholinesterase (AChE)-as a known target of toxins in fathead minnows and Daphnia magna, causing the inhibition of AChE-was analyzed. Finally, the structural aquatic toxicity alerts obtained from ToxAlert were used to confirm the possible mechanism of action. Machine learning and cheminformatics tools were used to analyze the data. Counter-propagation artificial neural network (CPANN) models were used to determine key binding properties of FDs to proteins associated with aquatic toxicity. Predicting the binding affinity of unknown FDs using quantitative structure-activity relationship (QSAR) models eliminates the need for complex and time-consuming calculations. The results of the study show which structural features of FDs have the greatest impact on aquatic organisms and help prioritize FDs and make manufacturing decisions.

DeepBindGCN: Integrating Molecular Vector Representation with Graph Convolutional Neural Networks for Protein-Ligand Interaction Prediction.

The core of large-scale drug virtual screening is to select the binders accurately and efficiently with high affinity from large libraries of small molecules in which non-binders are usually dominant. The binding affinity is significantly influenced by the protein pocket, ligand spatial information, and residue types/atom types. Here, we used the pocket residues or ligand atoms as the nodes and constructed edges with the neighboring information to comprehensively represent the protein pocket or ligand information. Moreover, the model with pre-trained molecular vectors performed better than the one-hot representation. The main advantage of DeepBindGCN is that it is independent of docking conformation, and concisely keeps the spatial information and physical-chemical features. Using TIPE3 and PD-L1 dimer as proof-of-concept examples, we proposed a screening pipeline integrating DeepBindGCN and other methods to identify strong-binding-affinity compounds. It is the first time a non-complex-dependent model has achieved a root mean square error (RMSE) value of 1.4190 and Pearson r value of 0.7584 in the PDBbind v.2016 core set, respectively, thereby showing a comparable prediction power with the state-of-the-art affinity prediction models that rely upon the 3D complex. DeepBindGCN provides a powerful tool to predict the protein-ligand interaction and can be used in many important large-scale virtual screening application scenarios.

Sfcnn: a novel scoring function based on 3D convolutional neural network for accurate and stable protein-ligand affinity prediction.

BACKGROUND: Computer-aided drug design provides an effective method of identifying lead compounds. However, success rates are significantly bottlenecked by the lack of accurate and reliable scoring functions needed to evaluate binding affinities of protein-ligand complexes. Therefore, many scoring functions based on machine learning or deep learning have been developed to improve prediction accuracies in recent years. In this work, we proposed a novel featurization method, generating a new scoring function model based on 3D convolutional neural network. RESULTS: This work showed the results from testing four architectures and three featurization methods, and outlined the development of a novel deep 3D convolutional neural network scoring function model. This model simplified feature engineering, and in combination with Grad-CAM made the intermediate layers of the neural network more interpretable. This model was evaluated and compared with other scoring functions on multiple independent datasets. The Pearson correlation coefficients between the predicted binding affinities by our model and the experimental data achieved 0.7928, 0.7946, 0.6758, and 0.6474 on CASF-2016 dataset, CASF-2013 dataset, CSAR_HiQ_NRC_set, and Astex_diverse_set, respectively. Overall, our model performed accurately and stably enough in the scoring power to predict the binding affinity of a protein-ligand complex. CONCLUSIONS: These results indicate our model is an excellent scoring function, and performs well in scoring power for accurately and stably predicting the protein-ligand affinity. Our model will contribute towards improving the success rate of virtual screening, thus will accelerate the development of potential drugs or novel biologically active lead compounds.

A reinforcement learning approach for protein-ligand binding pose prediction.

Protein ligand docking is an indispensable tool for computational prediction of protein functions and screening drug candidates. Despite significant progress over the past two decades, it is still a challenging problem, characterized by the still limited understanding of the energetics between proteins and ligands, and the vast conformational space that has to be searched to find a satisfactory solution. In this project, we developed a novel reinforcement learning (RL) approach, the asynchronous advantage actor-critic model (A3C), to address the protein ligand docking problem. The overall framework consists of two models. During the search process, the agent takes an action selected by the actor model based on the current location. The critic model then evaluates this action and predict the distance between the current location and true binding site. Experimental results showed that in both single- and multi-atom cases, our model improves binding site prediction substantially compared to a naïve model. For the single-atom ligand, copper ion (Cu2+), the model predicted binding sites have a median root-mean-square-deviation (RMSD) of 2.39 Å to the true binding sites when starting from random starting locations. For the multi-atom ligand, sulfate ion (SO42-), the predicted binding sites have a median RMSD of 3.82 Å to the true binding sites. The ligand-specific models built in this study can be used in solvent mapping studies and the RL framework can be readily scaled up to larger and more diverse sets of ligands.

Computational characterization of homologous ligands binding to a deep hydrophobic pocket in Shigella flexneri pilot protein MxiM.

The type III secretion system (T3SS) is an important molecular machinery in gram-negative bacteria Shigella flexneri as it provides ways for translocating virulence factors from the bacteria into host cells, eventually leading to severe disease symptoms such as bacillary dysentery. Due to the rising concerns of antibiotics resistance in bactericidal strategy, the anti-virulence strategy that primarily targets the T3SS components becomes an attractive alternative. MxiM, the secretin pilot protein of Shigella flexneri, binds the secretin MxiD and facilitates the formation of the secretin ring in outer membrane in T3SS assembly. MxiM harbors a large hydrophobic pocket that has been shown to be important in MxiM-MxiD interaction. In this work, I examined the ligand binding property of MxiM by performing molecular dynamics (MD) simulations of the association between MxiM and a series of hydrophobic ligands, with simulation time amounted to 30 µs. MD simulations successfully captured spontaneous ligand binding events in 153 of the 300 trajectories. The ligand binding can be categorized into two types: a fast type, in which the ligand binds quickly into the hydrophobic pocket and a slow type, in which the ligand forms an encounter complex with the protein before binding into the hydrophobic pocket. Using the MxiM-ligand binding poses captured in MD simulations, I additionally performed umbrella-sampling MD simulations with total simulation time amounted to 63 µs to obtain protein-ligand binding free energies. The relationship between the ligand binding free energy and ligand size appears to be nonlinear and exhibits an exponential decay pattern. In summary, I performed computational characterization of MxiM-hydrophobic ligand binding capabilities and properties, which may provide valuable insights into designing anti-bacterial medicine against antibiotics resistance in Shigella flexneri.