Structure-based Prediction Methods Research Articles

DiscoTope-3.0: improved B-cell epitope prediction using inverse folding latent representations.

Accurate computational identification of B-cell epitopes is crucial for the development of vaccines, therapies, and diagnostic tools. However, current structure-based prediction methods face limitations due to the dependency on experimentally solved structures. Here, we introduce DiscoTope-3.0, a markedly improved B-cell epitope prediction tool that innovatively employs inverse folding structure representations and a positive-unlabelled learning strategy, and is adapted for both solved and predicted structures. Our tool demonstrates a considerable improvement in performance over existing methods, accurately predicting linear and conformational epitopes across multiple independent datasets. Most notably, DiscoTope-3.0 maintains high predictive performance across solved, relaxed and predicted structures, alleviating the need for experimental structures and extending the general applicability of accurate B-cell epitope prediction by 3 orders of magnitude. DiscoTope-3.0 is made widely accessible on two web servers, processing over 100 structures per submission, and as a downloadable package. In addition, the servers interface with RCSB and AlphaFoldDB, facilitating large-scale prediction across over 200 million cataloged proteins. DiscoTope-3.0 is available at: https://services.healthtech.dtu.dk/service.php?DiscoTope-3.0.

DeepProSite: structure-aware protein binding site prediction using ESMFold and pretrained language model.

Identifying the functional sites of a protein, such as the binding sites of proteins, peptides, or other biological components, is crucial for understanding related biological processes and drug design. However, existing sequence-based methods have limited predictive accuracy, as they only consider sequence-adjacent contextual features and lack structural information. In this study, DeepProSite is presented as a new framework for identifying protein binding site that utilizes protein structure and sequence information. DeepProSite first generates protein structures from ESMFold and sequence representations from pretrained language models. It then uses Graph Transformer and formulates binding site predictions as graph node classifications. In predicting protein-protein/peptide binding sites, DeepProSite outperforms state-of-the-art sequence- and structure-based methods on most metrics. Moreover, DeepProSite maintains its performance when predicting unbound structures, in contrast to competing structure-based prediction methods. DeepProSite is also extended to the prediction of binding sites for nucleic acids and other ligands, verifying its generalization capability. Finally, an online server for predicting multiple types of residue is established as the implementation of the proposed DeepProSite. The datasets and source codes can be accessed at https://github.com/WeiLab-Biology/DeepProSite. The proposed DeepProSite can be accessed at https://inner.wei-group.net/DeepProSite/.

DG-Affinity: predicting antigen–antibody affinity with language models from sequences

BackgroundAntibody-mediated immune responses play a crucial role in the immune defense of human body. The evolution of bioengineering has led the progress of antibody-derived drugs, showing promising efficacy in cancer and autoimmune disease therapy. A critical step of this development process is obtaining the affinity between antibodies and their binding antigens.ResultsIn this study, we introduce a novel sequence-based antigen–antibody affinity prediction method, named DG-Affinity. DG-Affinity uses deep neural networks to efficiently and accurately predict the affinity between antibodies and antigens from sequences, without the need for structural information. The sequences of both the antigen and the antibody are first transformed into embedding vectors by two pre-trained language models, then these embeddings are concatenated into an ConvNeXt framework with a regression task. The results demonstrate the superiority of DG-Affinity over the existing structure-based prediction methods and the sequence-based tools, achieving a Pearson’s correlation of over 0.65 on an independent test dataset.ConclusionsCompared to the baseline methods, DG-Affinity achieves the best performance and can advance the development of antibody design. It is freely available as an easy-to-use web server at https://www.digitalgeneai.tech/solution/affinity.

Clustered tree regression to learn protein energy change with mutated amino acid.

Accurate and effective prediction of mutation-induced protein energy change remains a great challenge and of great interest in computational biology. However, high resource consumption and insufficient structural information of proteins severely limit the experimental techniques and structure-based prediction methods. Here, we design a structure-independent protocol to accurately and effectively predict the mutation-induced protein folding free energy change with only sequence, physicochemical and evolutionary features. The proposed clustered tree regression protocol is capable of effectively exploiting the inherent data patterns by integrating unsupervised feature clustering by K-means and supervised tree regression using XGBoost, and thus enabling fast and accurate protein predictions with different mutations, with an average Pearson correlation coefficient of 0.83 and an average root-mean-square error of 0.94kcal/mol. The proposed sequence-based method not only eliminates the dependence on protein structures, but also has potential applications in protein predictions with rare structural information.

Sequence-structure functional implications and molecular simulation of high deleterious nonsynonymous substitutions in IDH1 revealed the mechanism of drug resistance in glioma.

In the past few years, various somatic point mutations of isocitrate dehydrogenase (IDH) encoding genes (IDH1 and IDH2) have been identified in a broad range of cancers, including glioma. Despite the important function of IDH1 in tumorigenesis and its very polymorphic nature, it is not yet clear how different nsSNPs affect the structure and function of IDH1. In the present study, we employed different machine learning algorithms to screen nsSNPs in the IDH1 gene that are highly deleterious. From a total of 207 SNPs, all of the servers classified 80 mutations as deleterious. Among the 80 deleterious mutations, 14 were reported to be highly destabilizing using structure-based prediction methods. Three highly destabilizing mutations G15E, W92G, and I333S were further subjected to molecular docking and simulation validation. The docking results and molecular simulation analysis further displayed variation in dynamics features. The results from molecular docking and binding free energy demonstrated reduced binding of the drug in contrast to the wild type. This, consequently, shows the impact of these deleterious substitutions on the binding of the small molecule. PCA (principal component analysis) and FEL (free energy landscape) analysis revealed that these mutations had caused different arrangements to bind small molecules than the wild type where the total internal motion is decreased, thus consequently producing minimal binding effects. This study is the first extensive in silico analysis of the IDH1 gene that can narrow down the candidate mutations for further validation and targeting for therapeutic purposes.

Learning spatial structures of proteins improves protein–protein interaction prediction

Spatial structures of proteins are closely related to protein functions. Integrating protein structures improves the performance of protein-protein interaction (PPI) prediction. However, the limited quantity of known protein structures restricts the application of structure-based prediction methods. Utilizing the predicted protein structure information is a promising method to improve the performance of sequence-based prediction methods. We propose a novel end-to-end framework, TAGPPI, to predict PPIs using protein sequence alone. TAGPPI extracts multi-dimensional features by employing 1D convolution operation on protein sequences and graph learning method on contact maps constructed from AlphaFold. A contact map contains abundant spatial structure information, which is difficult to obtain from 1D sequence data directly. We further demonstrate that the spatial information learned from contact maps improves the ability of TAGPPI in PPI prediction tasks. We compare the performance of TAGPPI with those of nine state-of-the-art sequence-based methods, and TAGPPI outperforms such methods in all metrics. To the best of our knowledge, this is the first method to use the predicted protein topology structure graph for sequence-based PPI prediction. More importantly, our proposed architecture could be extended to other prediction tasks related to proteins.

Rational Prediction with Molecular Dynamics for Hit Identification

Although the motions of proteins are fundamental for their function, for pragmatic reasons, the consideration of protein elasticity has traditionally been neglected in drug discovery and design. This review details protein motion, its relevance to biomolecular interactions and how it can be sampled using molecular dynamics simulations. Within this context, two major areas of research in structure-based prediction that can benefit from considering protein flexibility, binding site detection and molecular docking, are discussed. Basic classification metrics and statistical analysis techniques, which can facilitate performance analysis, are also reviewed. With hardware and software advances, molecular dynamics in combination with traditional structure-based prediction methods can potentially reduce the time and costs involved in the hit identification pipeline.

On the Structural Context and Identification of Enzyme Catalytic Residues

Enzymes play important roles in most of the biological processes. Although only a small fraction of residues are directly involved in catalytic reactions, these catalytic residues are the most crucial parts in enzymes. The study of the fundamental and unique features of catalytic residues benefits the understanding of enzyme functions and catalytic mechanisms. In this work, we analyze the structural context of catalytic residues based on theoretical and experimental structure flexibility. The results show that catalytic residues have distinct structural features and context. Their neighboring residues, whether sequence or structure neighbors within specific range, are usually structurally more rigid than those of noncatalytic residues. The structural context feature is combined with support vector machine to identify catalytic residues from enzyme structure. The prediction results are better or comparable to those of recent structure-based prediction methods.

Rational prediction with molecular dynamics for hit identification.

Although the motions of proteins are fundamental for their function, for pragmatic reasons, the consideration of protein elasticity has traditionally been neglected in drug discovery and design. This review details protein motion, its relevance to biomolecular interactions and how it can be sampled using molecular dynamics simulations. Within this context, two major areas of research in structure-based prediction that can benefit from considering protein flexibility, binding site detection and molecular docking, are discussed. Basic classification metrics and statistical analysis techniques, which can facilitate performance analysis, are also reviewed. With hardware and software advances, molecular dynamics in combination with traditional structure-based prediction methods can potentially reduce the time and costs involved in the hit identification pipeline.

Predicting Peptide Binding Affinities to MHC Molecules Using a Modified Semi-Empirical Scoring Function

The Major Histocompatibility Complex (MHC) plays an important role in the human immune system. The MHC is involved in the antigen presentation system assisting T cells to identify foreign or pathogenic proteins. However, an MHC molecule binding a self-peptide may incorrectly trigger an immune response and cause an autoimmune disease, such as multiple sclerosis. Understanding the molecular mechanism of this process will greatly assist in determining the aetiology of various diseases and in the design of effective drugs. In the present study, we have used the Fresno semi-empirical scoring function and modify the approach to the prediction of peptide-MHC binding by using open-source and public domain software. We apply the method to HLA class II alleles DR15, DR1, and DR4, and the HLA class I allele HLA A2. Our analysis shows that using a large set of binding data and multiple crystal structures improves the predictive capability of the method. The performance of the method is also shown to be correlated to the structural similarity of the crystal structures used. We have exposed some of the obstacles faced by structure-based prediction methods and proposed possible solutions to those obstacles. It is envisaged that these obstacles need to be addressed before the performance of structure-based methods can be on par with the sequence-based methods.

Support vector machine prediction of enzyme function with conjoint triad feature and hierarchical context

BackgroundEnzymes are known as the largest class of proteins and their functions are usually annotated by the Enzyme Commission (EC), which uses a hierarchy structure, i.e., four numbers separated by periods, to classify the function of enzymes. Automatically categorizing enzyme into the EC hierarchy is crucial to understand its specific molecular mechanism.ResultsIn this paper, we introduce two key improvements in predicting enzyme function within the machine learning framework. One is to introduce the efficient sequence encoding methods for representing given proteins. The second one is to develop a structure-based prediction method with low computational complexity. In particular, we propose to use the conjoint triad feature (CTF) to represent the given protein sequences by considering not only the composition of amino acids but also the neighbor relationships in the sequence. Then we develop a support vector machine (SVM)-based method, named as SVMHL (SVM for hierarchy labels), to output enzyme function by fully considering the hierarchical structure of EC. The experimental results show that our SVMHL with the CTF outperforms SVMHL with the amino acid composition (AAC) feature both in predictive accuracy and Matthew’s correlation coefficient (MCC). In addition, SVMHL with the CTF obtains the accuracy and MCC ranging from 81% to 98% and 0.82 to 0.98 when predicting the first three EC digits on a low-homologous enzyme dataset. We further demonstrate that our method outperforms the methods which do not take account of hierarchical relationship among enzyme categories and alternative methods which incorporate prior knowledge about inter-class relationships.ConclusionsOur structure-based prediction model, SVMHL with the CTF, reduces the computational complexity and outperforms the alternative approaches in enzyme function prediction. Therefore our new method will be a useful tool for enzyme function prediction community.

The use of binding-prediction models to identify M. bovis-specific antigenic peptides for screening assays in bovine tuberculosis

The identification of MHC class II-restricted antigenic peptides for inclusion into vaccines and/or as diagnostic test reagents for mycobacterial infections remains a high research priority. To expedite discovery of such peptides, numerous bioinformatic tools have been developed to predict whether a given peptide is likely to form a stable binding interaction with MHC class II molecules. However, no prediction tool dedicated to the identification of bovine MHC (BoLA) class II-restricted peptides is currently available. Using experimental immunogenicity data derived from the stimulation of whole blood of Mycobacterium bovis-infected cattle with 105 individual M. bovis-derived peptides, we have compared the ability of a novel BoLA DRB3 structure-based prediction method (Hepitom) with the human MHC class II binding predictor model ProPred in predicting peptides that induce bovine T-cell activation. When a stringent cut off for considering peptide antigenicity was applied, the sensitivities of Hepitom and ProPred in detecting immunogenic peptides were 62% and 77%, respectively. In contrast, the Hepitom model showed greater specificity, with values of 66% and 34% for Hepitom and ProPred, respectively. Using all peptides, seven out of eleven M. bovis proteins were identified as being highly immunogenic. All but one of these antigens were also identified when just the Hepitom predicted peptides were used, while only four of the seven were identified using the ProPred predicted peptides. In conclusion, we demonstrate that the Hepitom model is a useful pre-screening tool to select peptides for further immunogenicity studies in cattle without major impact on the identification of antigenic M. bovis proteins.

Towards Universal Structure-Based Prediction of Class II MHC Epitopes for Diverse Allotypes

The binding of peptide fragments of antigens to class II MHC proteins is a crucial step in initiating a helper T cell immune response. The discovery of these peptide epitopes is important for understanding the normal immune response and its misregulation in autoimmunity and allergies and also for vaccine design. In spite of their biomedical importance, the high diversity of class II MHC proteins combined with the large number of possible peptide sequences make comprehensive experimental determination of epitopes for all MHC allotypes infeasible. Computational methods can address this need by predicting epitopes for a particular MHC allotype. We present a structure-based method for predicting class II epitopes that combines molecular mechanics docking of a fully flexible peptide into the MHC binding cleft followed by binding affinity prediction using a machine learning classifier trained on interaction energy components calculated from the docking solution. Although the primary advantage of structure-based prediction methods over the commonly employed sequence-based methods is their applicability to essentially any MHC allotype, this has not yet been convincingly demonstrated. In order to test the transferability of the prediction method to different MHC proteins, we trained the scoring method on binding data for DRB1*0101 and used it to make predictions for multiple MHC allotypes with distinct peptide binding specificities including representatives from the other human class II MHC loci, HLA-DP and HLA-DQ, as well as for two murine allotypes. The results showed that the prediction method was able to achieve significant discrimination between epitope and non-epitope peptides for all MHC allotypes examined, based on AUC values in the range 0.632–0.821. We also discuss how accounting for peptide binding in multiple registers to class II MHC largely explains the systematically worse performance of prediction methods for class II MHC compared with those for class I MHC based on quantitative prediction performance estimates for peptide binding to class II MHC in a fixed register.

Composition-based effective chain length for prediction of protein folding rates

Folding rate prediction is a useful way to find the key factors affecting folding kinetics of proteins. Structural information is more or less required in the present prediction methods, which limits the application of these methods to various proteins. In this work, an "effective length" is defined solely based on the composition of a protein, namely, the number of specific types of amino acids in a protein. A physical theory based on a minimalist model is employed to describe the relation between the folding rates and the effective length of proteins. Based on the resultant relationship between folding rates and effective length, the optimal sets of amino acids are found through the enumeration over all possible combinations of amino acids. This optimal set achieves a high correlation (with the coefficient of 0.84) between the folding rates and the optimal effective length. The features of these amino acids are consistent with our model and landscape theory. Further comparisons between our effective length and other factors are carried out. The effective length is physically consistent with structure-based prediction methods and has the best predictability for folding rates. These results all suggest that both entropy and energetics contribute importantly to folding kinetics. The ability to accurately and efficiently predict folding rates from composition enables the analysis of the kinetics for various kinds of proteins. The underlying physics in our method may be helpful to stimulate further understanding on the effects of various amino acids in folding dynamics.

Antimalarial Drug Discovery: In Silico Structural Biology and Rational Drug Design

Malaria remains one of the most burdensome human infectious diseases, with a high rate of resistance outbreaks and a constant need for the discovery of novel antimalarials and drug targets. For several reasons, Plasmodial proteins are difficult to characterise structurally using traditional physical approaches. However, these problems can be partially overcome using a number of in silico approaches. This review describes the peculiarities of malaria proteins and then details various in silico strategies to select and allow descriptions of the molecular structures of drug target candidates as well as subsequent rational approaches for drug design. Chiefly, homology modelling with specific focus on unique aspects of malaria proteins including low homology, large protein size and the presence of parasite-specific inserts is addressed and alternative strategies including multiple sequence and structure-based prediction methods, sampling-based approaches that aim to reveal likely global or shared features of a Plasmodial structure and the value of molecular dynamics understanding of unique features of Plasmodial proteins are discussed. Once a detailed description of the drug target is available, in silico approaches to the specific design of an inhibitory drug thereof becomes invaluable as an economic and rational alternative to chemical library screening.

Structure-based prediction of transcription factor binding sites using a protein-DNA docking approach

Accurate identification of transcription factor binding sites is critical to our understanding of transcriptional regulatory networks. To overcome the issue of high false-positive predictions that trouble the sequence-based prediction techniques, we have developed a structure-based prediction method that takes into consideration of interactions between the amino acids of a transcription factor and the nucleotides of its DNA binding sequence at structural level, along with an efficient protein-DNA docking algorithm. The docked structures between a protein and a DNA are evaluated using a knowledge-based energy function, in conjunction with van der Waals energy. Our docking algorithm supports quasi-flexible docking, overcoming a number of limiting issues faced by similar docking algorithms. Our rigid-body docking algorithm is tested on a dataset of 141 nonredundant transcription factor-DNA complex structures. The test results show that 63.1% of the 141 complex structures are reconstructed with accuracies better than 1.0 A RMSDs (root mean square deviation) and 79.4% of the complexes are predicted with accuracies better than 3.0 A RMSDs when using the native DNA structures. Our quasi-flexible docking algorithm, assuming that the DNA structures are not known, is tested on a separate set of 45 transcription factor-DNA complexes, of which 57.8% of the docked complex conformations achieve better than 1.0 A RMSDs while 71.1% of the complexes have RMSDs less than 3.0 A. We have also applied our method to predict the binding motifs of the ferric uptake regulator in E. coli and showed that most of the experimentally identified sites can be predicted accurately.

DynaPred: A structure and sequence based method for the prediction of MHC class I binding peptide sequences and conformations

The binding of endogenous antigenic peptides to MHC class I molecules is an important step during the immunologic response of a host against a pathogen. Thus, various sequence- and structure-based prediction methods have been proposed for this purpose. The sequence-based methods are computationally efficient, but are hampered by the need of sufficient experimental data and do not provide a structural interpretation of their results. The structural methods are data-independent, but are quite time-consuming and thus not suited for screening of whole genomes. Here, we present a new method, which performs sequence-based prediction by incorporating information obtained from molecular modeling. This allows us to perform large databases screening and to provide structural information of the results. We developed a SVM-trained, quantitative matrix-based method for the prediction of MHC class I binding peptides, in which the features of the scoring matrix are energy terms retrieved from molecular dynamics simulations. At the same time we used the equilibrated structures obtained from the same simulations in a simple and efficient docking procedure. Our method consists of two steps: First, we predict potential binders from sequence data alone and second, we construct protein-peptide complexes for the predicted binders. So far, we tested our approach on the HLA-A0201 allele. We constructed two prediction models, using local, position-dependent (DynaPred(POS)) and global, position-independent (DynaPred) features. The former model outperformed the two sequence-based methods used in our evaluation; the latter shows a much higher generalizability towards other alleles than the position-dependent models. The constructed peptide structures can be refined within seconds to structures with an average backbone RMSD of 1.53 A from the corresponding experimental structures.