Evaluation of AlphaFold antibody-antigen modeling with implications for improving predictive accuracy

AbEpiTope-1.0: Improved antibody target prediction by use of AlphaFold and inverse folding

B cell epitope prediction tools are crucial for designing vaccines and disease diagnostics. However, predicting which antigens a specific antibody binds to and their exact binding sites (epitopes) remains challenging. Here, we present AbEpiTope-1.0, a tool for antibody-specific B cell epitope prediction, using AlphaFold for structural modeling and inverse folding for machine learning models. On a dataset of 1730 antibody-antigen complexes, AbEpiTope-1.0 outperforms AlphaFold in predicting modeled antibody-antigen interface accuracy. By creating swapped antibody-antigen complex structures for each antibody-antigen complex using incorrect antibodies, we show that predicted accuracies are sensitive to antibody input. Furthermore, a model variant optimized for antibody target prediction-differentiating true from swapped complexes-achieved an accuracy of 61.21% in correctly identifying antibody-antigen pairs. The tool evaluates hundreds of structures in minutes, providing researchers with a resource for screening antibodies targeting specific antigens. AbEpiTope-1.0 is freely available as a web server and software.

Cryo-EM structures of HCV E2 glycoprotein bound to neutralizing and non-neutralizing antibodies determined using bivalent Fabs as fiducial markers

Global elimination of hepatitis C virus (HCV) will require an effective cross-genotype vaccine. The HCV E2 envelope glycoprotein is the main target of neutralizing antibodies but also contains epitopes that elicit non-neutralizing antibodies which may provide protection through Fc effector functions rather than direct neutralization. We determined cryo-EM structures of a broadly neutralizing antibody, a moderately neutralizing antibody, and a non-neutralizing antibody bound to E2 to resolutions of 3.8, 3.3, and 3.7 Å, respectively. Whereas the broadly neutralizing antibody targeted the front layer of E2 and the non-neutralizing antibody targeted the back layer, the moderately neutralizing antibody straddled both front and back layers, and thereby defined a new neutralizing epitope on E2. The small size of complexes between conventional (monovalent) Fabs and E2 (~110 kDa) presented a challenge for cryo-EM. Accordingly, we engineered bivalent versions of E2-specific Fabs that doubled the size of Fab-E2 complexes and conferred highly identifiable shapes to the complexes that facilitated particle selection and orientation for image processing. This study validates bivalent Fabs as new fiducial markers for cryo-EM analysis of small proteins such as HCV E2 and identifies a new target epitope for vaccine development.

Machine learning to predict de novo protein-protein interactions

Advances in machine learning for structural biology have dramatically enhanced our capacity to predict protein-protein interactions (PPIs). Here, we review recent developments in the computational prediction of PPIs, particularly focusing on innovations that enable interaction predictions that have no precedence in nature, termed de novo. We discuss novel machine learning algorithms for PPI prediction, including approaches based on co-folding and atomic graphs. We further highlight methods that learn from molecular surfaces, which can predict PPIs not found in nature including interactions induced by small molecules. Finally, we explore the emerging biotechnological applications enabled by these predictive capabilities, including the prediction of antibody-antigen complexes and molecular glue-induced PPIs, and discuss their potential to empower drug discovery and protein engineering.

Evaluation of Alphafold modeling for elucidation of nanobody-peptide epitope interactions

Models of Ab-antigen complexes can be used to understand interaction mechanisms and for improving affinity. This study evaluates the use of the protein structure prediction algorithm AlphaFold (AF) for exploration of interactions between peptide epitope tags and the smallest functional antibody fragments, nanobodies (Nbs). Although past studies of AF for modeling antibody-target (antigen) interactions suggested modest algorithm performance, those were primarily focused on Ab-protein interactions, while the performance and utility of AF for Nb-peptide interactions, which are generally less complex due to smaller antigens, smaller binding domains, and fewer chains, is less clear. In this study we evaluated the performance of AF for predicting the structures of Nbs bound to experimentally validated, linear, short peptide epitopes (Nb-tag pairs). We expanded the pool of experimental data available for comparison through crystallization and structural determination of a previously reported Nb-tag complex (Nb127). Models of Nb-tag pair structures generated from AF were variable with respect to consistency with experimental data, with good performance in just over half (4 out of 6) of cases. Even among Nb-tag pairs successfully modeled in isolation, efforts to translate modeling to more complex contexts failed, suggesting an underappreciated role of the size and complexity of inputs in AF modeling success. Finally, the model of a Nb-tag pair with minimal previous characterization was used to guide the design of a peptide-electrophile conjugate that undergoes covalent crosslinking with Nb upon binding. These findings highlight the utility of minimized antibody and antigen structures to maximize insights from AF modeling.

Unveiling the influence of fastest nobel prize winner discovery: alphafold's algorithmic intelligence in medical sciences

CONTEXT

AlphaFold's advanced AI technology has transformed protein structure interpretation. By predicting three-dimensional protein structures from amino acid sequences, AlphaFold has solved the complex protein-folding problem, previously challenging for experimental methods due to numerous possible conformations. Since its inception, AlphaFold has introduced several versions, including AlphaFold2, AlphaFold DB, AlphaFold Multimer, Alpha Missense, and AlphaFold3, each further enhancing protein structure prediction. Remarkably, AlphaFold is recognized as the fastest Nobel Prize winner in science history. This technology has extensive applications, potentially transforming treatment and diagnosis in medical sciences by reducing drug design costs and time, while elucidating structural pathways of human body systems. Numerous studies have demonstrated how AlphaFold aids in understanding health conditions by providing critical information about protein mutations, abnormal protein-protein interactions, and changes in protein dynamics. Researchers have also developed new technologies and pipelines using different versions of AlphaFold to amplify its potential. However, addressing existing limitations is crucial to maximizing AlphaFold's capacity to redefine medical research. This article reviews AlphaFold's impact on five key aspects of medical sciences: protein mutation, protein-protein interaction, molecular dynamics, drug design, and immunotherapy.

METHODS

This review examines the contributions of various AlphaFold versions AlphaFold2, AlphaFold DB, AlphaFold Multimer, Alpha Missense, and AlphaFold3 to protein structure prediction. The methods include an extensive analysis of computational techniques and software used in interpreting and predicting protein structures, emphasizing advances in AI technology and its applications in medical research.

Recent progress and future challenges in structure-based protein-protein interaction prediction

Protein-protein interactions (PPIs) play a fundamental role in cellular processes, and understanding these interactions is crucial for advances in both basic biological science and biomedical applications. This review presents an overview of recent progress in computational methods for modeling protein complexes and predicting PPIs based on 3D structures, focusing on the transformative role of artificial intelligence-based approaches. We further discuss the expanding biomedical applications of PPI research, including the elucidation of disease mechanisms, drug discovery, and therapeutic design. Despite these advances, significant challenges remain in predicting host-pathogen interactions, interactions between intrinsically disordered regions, and interactions related to immune responses. These challenges are worthwhile for future explorations and represent the frontier of research in this field.

Antibody immunotherapies for personalized opioid addiction treatment

Approved therapies for managing opioid addiction involve intensive treatment regimens which remain both costly and ineffective. As pharmaceutical interventions have achieved variable success treating substance use disorders (SUD), alternative therapeutics must be considered. Antidrug antibodies induced by vaccination or introduced as monoclonal antibody formulations can neutralize or destroy opioids in circulation before they reach their central nervous system targets or act as enzymes to deactivate opioid receptors, preventing the physiologic and psychoactive effects of the substance. A lack of "reward" for those suffering from SUD has been shown to result in cessation of use and promote long-term abstinence. Decreased antibody production costs and the advent of novel gene therapies that stimulate in vivo production of monoclonal antibodies have renewed interest in this strategy. Furthermore, advances in understanding of SUD immunopathogenesis have revealed distinct mechanisms of neuroimmune dysregulation underlying the disorder. Beyond assisting with cessation of drug use, antibody therapies could treat or reverse pathophysiologic hallmarks that contribute to addiction and which could be the cause of chronic cognitive defects resulting from drug use. In this review, we synthesize key current literature regarding the efficacy of immunotherapies in managing opioid addiction and SUD. We will explore the neuropharmacology underlying these treatments by relating evidence from studies on the use of antibody therapeutics to counteract various drug behaviors and by drawing parallels to the similar immunopathology observed in neurodegenerative disorders. Finally, we will discuss the implications of novel immunization technologies and the application of computational methods in developing personalized addiction treatments. SIGNIFICANCE STATEMENT: Significant new evidence contributing to our understanding of substance use disorders has recently emerged leading to a paradigm shift concerning the role of immunology in the neuropathogenesis of opioid use disorder. Concurrently, immunotherapeutic technologies such as antibody therapeutics have advanced the capabilities regarding applications that take advantage of these key principles. This article reviews key antibody-based treatments being studied and highlights directions for further research that may contribute to the management of opioid use disorder.

In silico prediction method for plant Nucleotide-binding leucine-rich repeat- and pathogen effector interactions

Plant Nucleotide-binding leucine-rich repeat (NLR) proteins play a crucial role in effector recognition and activation of Effector triggered immunity following pathogen infection. Genome sequencing advancements have led to the identification of a myriad of NLRs in numerous agriculturally important plant species. However, deciphering which NLRs recognize specific pathogen effectors remains challenging. Predicting NLR-effector interactions in silico will provide a more targeted approach for experimental validation, critical for elucidating function, and advancing our understanding of NLR-triggered immunity. In this study, NLR-effector protein complex structures were predicted using AlphaFold2-Multimer for all experimentally validated NLR-effector interactions reported in literature. Binding affinities- and energies were predicted using 97 machine learning models from Area-Affinity. We show that AlphaFold2-Multimer predicted structures have acceptable accuracy and can be used to investigate NLR-effector interactions in silico. Binding affinities for 58 NLR-effector complexes ranged between -8.5 and -10.6 log(K), and binding energies between -11.8 and -14.4 kcal/mol-1, depending on the Area-Affinity model used. For 2427 "forced" NLR-effector complexes, these estimates showed larger variability, enabling identification of novel NLR-effector interactions with 99% accuracy using an Ensemble machine learning model. The narrow range of binding energies- and affinities for "true" interactions suggest a specific change in Gibbs free energy, and thus conformational change, is required for NLR activation. This is the first study to provide a method for predicting NLR-effector interactions, applicable to all pathosystems. Finally, the NLR-Effector Interaction Classification (NEIC) resource can streamline research efforts by identifying NLRs important for plant-pathogen resistance, advancing our understanding of plant immunity.

Identification and functional validation of a novel FBN1 variant in a Marfan syndrome family using a zebrafish model

BACKGROUND

Marfan syndrome (MFS) is an inherited autosomal dominant disorder that affects connective tissue with an incidence of about 1 in 5,000 to 10,000 people. 90% of MFS is caused by mutations in the fibrillin-1 (FBN1) gene. We recruited a family with MFS phenotype in South China and identified a novel variant. This study investigated whether this genetic variant is pathogenic and the potential pathway related to lipid metabolism in MFS.

METHODS

A three-generation consanguineous family was recruited for this study. Whole exome sequencing (WES) was utilized on family members. The 3D structure of the protein was predicted using AlphaFold. CRISPR/Cas9 was applied to generate a similar fbn1 nonsense mutation (fbn1+/-) in zebrafish. RNA-seq analysis on zebrafish was performed to identify potential pathways related to MFS pathogenesis.

RESULTS

Our study identified a novel variant [NM_000138.5; c.7764 C > G: p.(Y2588*)] in FBN1 gene from the family and identified the same site mutation among the proband along with her son and daughter. Structural modeling showed the p.Y2588* mutation resulted from a truncated protein. Compared to wild-type zebrafish, the F2 generation fbn1+/- zebrafish exhibited MFS phenotype. RNA-seq analysis indicated that many genes related to leptin are up-regulating, which could affect bone development and adipose homeostasis.

CONCLUSION

A novel variant was identified in FBN1 gene. In a zebrafish model, we found functional evidence supporting the pathogenicity of the detected nonsense mutation. Our research proposes a possible mechanism underlying the relationship between lipid metabolism and MFS. These findings can help improve the clinical diagnosis and treatment of MFS.

Leveraging large language models to predict antibody biological activity against influenza A hemagglutinin

Monoclonal antibodies (mAbs) represent one of the most prevalent FDA-approved treatments for autoimmune, infectious, and cancer diseases. However, their discovery and development remains a time-consuming and costly process. Recent advancements in machine learning (ML) and artificial intelligence (AI) have shown significant promise in revolutionizing antibody discovery field. Models that predict antibody biological activity enable in silico evaluation of binding and functional properties; such models can prioritize antibodies with the highest likelihood of success in laboratory testing procedures. We explore an AI model for predicting the binding and receptor blocking activity of antibodies against influenza A hemagglutinin (HA) antigens. Our model is developed with the Molecular Aligned Multi-Modal Architecture and Language (MAMMAL) framework for biologics discovery to predict antibody-antigen interactions using only sequence information. To evaluate the model's performance, we tested it under various data split conditions to mimic real-world scenarios. Our model achieved an area under the receiver operating characteristic (AUROC) score of ≥ 0.91 for predicting the activity of existing antibodies against seen HAs and an AUROC score of 0.9 for unseen HAs. For novel antibody activity prediction, the AUROC was 0.73, which further declined to 0.63-0.66 under stringent constraints on similarity to existing antibodies. These results demonstrate the potential of AI foundation models to transform antibody design by reducing dependence on extensive laboratory testing and enabling more efficient prioritization of antibody candidates. Moreover, our findings emphasize the critical importance of diverse and comprehensive antibody datasets to improve the generalization of prediction models, particularly for novel antibody development.

RosettaHDX: Predicting antibody-antigen interaction from hydrogen-deuterium exchange mass spectrometry data

High-throughput characterization of antibody-antigen complexes at the atomic level is critical for understanding antibody function and enabling therapeutic development. Hydrogen-deuterium exchange mass spectrometry (HDX-MS) enables rapid epitope mapping, but its data are too sparse for independent structure determination. In this study, we introduce RosettaHDX, a hybrid method that combines computational docking with differential HDX-MS data to enhance the accuracy of antibody-antigen complex models beyond what either method can achieve individually. By incorporating HDX data as both distance restraints and a scoring term in the RosettaDock algorithm, RosettaHDX successfully generated near-native models (interface root-mean square deviation ≤ 4 Å) for all 9 benchmark complexes examined, averaging 3.6 times more near-native models than Rosetta alone. Near-native models among the top 10 scoring were identified in 3/9 cases, compared to 1/9 with Rosetta alone. Additionally, we developed a predictive metric based on docking results with HDX restraints to identify allosteric peptides in HDX datasets.

Rēs ipSAE loquunt: What's wrong with AlphaFold's ipTM score and how to fix it

AlphaFold's ipTM metric is used to predict the accuracy of structural predictions of protein-protein interactions (PPIs) and the probability that two proteins interact. Many AF2/AF3 users have experienced the phenomenon that if they trim full-length sequence constructs (e.g. from UniProt) to the interacting domains (or domain+peptide), their ipTM scores go up, even though the structure prediction of the interaction is unchanged. The reason this happens is due to the mathematical formulation of ipTM in AF2/AF3, which scores the interactions of whole chains. If both chains in a PPI complex contain large amounts of disorder or accessory domains that do not form the primary domain-domain or domain/peptide interaction, the ipTM score can be lowered significantly. The score then does not accurately represent the accuracy of the structure prediction nor whether the two proteins actually interact. We have solved this problem by: 1) including only residue pairs in the ipTM metric that have good predicted aligned error ( PAE ) scores; 2) by adjusting thed 0 parameter (a function of the length of the query sequences) in the TM score equation to include only the number of residues with good interchain PAE s to the aligned residue; and 3) using the PAE value itself and not the probability distributions over the aligned error to calculate the pairwise residue-residue p T M values that go into the ipTM calculation. The first two are crucial in calculating high ipTM s for domain-domain and domain-peptide interactions even in the presence of many hundreds of residues in disordered regions and/or accessory domains. The third allows us to require only the common output json files of AF2 and AF3 (including the server output) without having to change the AlphaFold code and without affecting the accuracy. We show in a benchmark that the new score, called ipSAE (interaction prediction Score from Aligned Errors), is able to separate true from false complexes more efficiently than AlphaFold2's ipTM score. The resulting program is freely available at https://github.com/dunbracklab/IPSAE.

Epitope Mapping with Sidewinder: An XL-MS and Structural Modeling Approach

Antibodies are critical to the host's immune defense against bacterial pathogens. Understanding the mechanisms of antibody-antigen interactions is essential for developing new targeted immunotherapies. Building computational workflows that can identify where an antibody binds its cognate antigen and deconvoluting the interaction interface in a high-throughput manner are critical for advancing this field. Cross-linking mass spectrometry (XL-MS) integrated with structural modeling offers a flexible and high-resolution strategy to map protein-protein interactions from low sample amounts. However, cross-linking and in silico modeling have limitations that require robust analytical workflows to make accurate inferences. In this study, we introduce Sidewinder, a modular high-throughput pipeline combining state-of-the-art computational structural prediction and molecular docking with rapid XL-MS analysis, enabling comprehensive interrogation of antibody-antigen systems. We validated this pipeline on antibodies targeting two Streptococcus pyogenes virulence factors. Using recently published data, we identified a well-defined monoclonal antibody epitope on Streptolysin O by generating and querying a large ensemble of interaction models probabilistically. We also showcased the utility of the Sidewinder pipeline by analyzing a more complex system, involving monoclonal antibodies that target the cell wall-anchored M1 protein. The flexibility and robustness of the Sidewinder pipeline provide a powerful framework for future studies of complex antibody-antigen systems, potentially leading to new therapeutic strategies.

Unveiling the new chapter in nanobody engineering: advances in traditional construction and AI-driven optimization

Nanobodies (Nbs), miniature antibodies consisting solely of the variable region of heavy chains, exhibit unique properties such as small size, high stability, and strong specificity, making them highly promising for disease diagnosis and treatment. The engineering production of Nbs has evolved into a mature process, involving library construction, screening, and expression purification. Different library types, including immune, naïve, and synthetic/semi-synthetic libraries, offer diverse options for various applications, while display platforms like phage display, cell surface display, and non-surface display provide efficient screening of target Nbs. Recent advancements in artificial intelligence (AI) have opened new avenues in Nb engineering. AI's exceptional performance in protein structure prediction and molecular interaction simulation has introduced novel perspectives and tools for Nb design and optimization. Integrating AI with traditional experimental methods is anticipated to enhance the efficiency and precision of Nb development, expediting the transition from basic research to clinical applications. This review comprehensively examines the latest progress in Nb engineering, emphasizing library construction strategies, display platform technologies, and AI applications. It evaluates the strengths and weaknesses of various libraries and display platforms and explores the potential and challenges of AI in predicting Nb structure, antigen-antibody interactions, and optimizing physicochemical properties.

Nanobody engineering: computational modelling and design for biomedical and therapeutic applications

Nanobodies, the smallest functional antibody fragment derived from camelid heavy-chain-only antibodies, have emerged as powerful tools for diverse biomedical applications. In this comprehensive review, we discuss the structural characteristics, functional properties, and computational approaches driving the design and optimisation of synthetic nanobodies. We explore their unique antigen-binding domains, highlighting the critical role of complementarity-determining regions in target recognition and specificity. This review further underscores the advantages of nanobodies over conventional antibodies from a biosynthesis perspective, including their small size, stability, and solubility, which make them ideal candidates for economical antigen capture in diagnostics, therapeutics, and biosensing. We discuss the recent advancements in computational methods for nanobody modelling, epitope prediction, and affinity maturation, shedding light on their intricate antigen-binding mechanisms and conformational dynamics. Finally, we examine a direct example of how computational design strategies were implemented for improving a nanobody-based immunosensor, known as a Quenchbody. Through combining experimental findings and computational insights, this review elucidates the transformative impact of nanobodies in biotechnology and biomedical research, offering a roadmap for future advancements and applications in healthcare and diagnostics.

An outlook on structural biology after AlphaFold: tools, limits and perspectives

AlphaFold and similar groundbreaking, AI-based tools, have revolutionized the field of structural bioinformatics, with their remarkable accuracy in ab-initio protein structure prediction. This success has catalyzed the development of new software and pipelines aimed at incorporating AlphaFold's predictions, often focusing on addressing the algorithm's remaining challenges. Here, we present the current landscape of structural bioinformatics shaped by AlphaFold, and discuss how the field is dynamically responding to this revolution, with new software, methods, and pipelines. While the excitement around AI-based tools led to their widespread application, it is essential to acknowledge that their practical success hinges on their integration into established protocols within structural bioinformatics, often neglected in the context of AI-driven advancements. Indeed, user-driven intervention is still as pivotal in the structure prediction process as in complementing state-of-the-art algorithms with functional and biological knowledge.

AlphaFold and what is next: bridging functional, systems and structural biology

INTRODUCTION

The DeepMind's AlphaFold (AF) has revolutionized biomedical and biocience research by providing both experts and non-experts with an invaluable tool for predicting protein structures. However, while AF is highly effective for predicting structures of rigid and globular proteins, it is not able to fully capture the dynamics, conformational variability, and interactions of proteins with ligands and other biomacromolecules.

AREAS COVERED

In this review, we present a comprehensive overview of the latest advancements in 3D model predictions for biomacromolecules using AF. We also provide a detailed analysis its of strengths and limitations, and explore more recent iterations, modifications, and practical applications of this strategy. Moreover, we map the path forward for expanding the landscape of AF toward predicting structures of every protein and peptide, and their interactions in the proteome in the most physiologically relevant form. This discussion is based on an extensive literature search performed using PubMed and Google Scholar.

EXPERT OPINION

While significant progress has been made to enhance AF's modeling capabilities, we argue that a combined approach integrating both various in silico and in vitro methods will be most beneficial for the future of structural biology, bridging the gaps between static and dynamic features of proteins and their functions.

What does AlphaFold3 learn about antigen and nanobody docking, and what remains unsolved?

Antibody therapeutic development is a major focus in healthcare. To accelerate drug development, significant efforts have been directed towards the in silico design and screening of antibodies for which high modeling accuracy is necessary. To probe AlphaFold3's (AF3) capabilities and limitations, we tested AF3's ability to capture the fine details and interplay between antibody structure prediction and antigen docking accuracy. With one seed, AF3 achieves an 11.0% and 11.4% high-accuracy docking success rate for antibodies and nanobodies, respectively, and a median unbound CDR H3 RMSD accuracy of 2.73 Å and 2.30 Å. CDR H3 accuracy boosts complex prediction accuracy, with antigen context improving CDR H3 accuracy, particularly for loops longer than 15 residues. Combining I-pLDDT with Δ G B improves discriminative power for correctly docked complexes. However, AF3's 60% failure rate for antibody and nanobody docking (with single seed sampling) demonstrates necessary refinement to improve antibody design endeavors.

Massive Sampling Strategy for Antibody-Antigen Targets in CAPRI Round 55 With MassiveFold

Massive sampling with AlphaFold2 improves protein-protein complex predictions. This has been shown during the last CASP15-CAPRI blind prediction round by the AFsample tool. However, more difficult targets including antibody-antigen binding remain challenging. CAPRI Round 55 consisted of three antibody-antigen targets and one heterotrimer. We used our AlphaFold2-based MassiveFold, running 6 prediction pools, each with their own set of parameters, to produce in total more than 6000 predictions per target. We show here that massive sampling categorically produces acceptable to high quality predictions, however it is clear that the AlphaFold2 confidence score cannot be used to identify the best models in the set. We also show that, contrary to what was done before for CASP15-CAPRI with AFsample, increasing the sampling without activating the dropout provides the best models for most of the targets of Round 55.

AlphaFold and Docking Approaches for Antibody-Antigen and Other Targets: Insights From CAPRI Rounds 47-55

Accurate modeling of the structures of protein-protein complexes and other biomolecular interactions represents a longstanding and important challenge for computational biology. The Critical Assessment of PRedicted Interactions (CAPRI) experiment has served for over two decades as a key means to assess and compare current approaches and methods through blind predictive scenarios, highlighting useful strategies, and new developments. Here we describe the performance of our laboratory's team in recent CAPRI rounds, which included submissions for 10 modeling targets. Our team utilized a range of docking and modeling approaches, including ZDOCK, Rosetta, and ZRANK, to model, refine, and score protein-protein and protein-DNA complexes. For recent targets we utilized adaptations of AlphaFold to generate models, leading to near-native models for an antibody-peptide target, and a highly accurate (but low ranked) model for an antibody-MHC complex. These results underscore the utility of AlphaFold-based protocols for predictive protein complex modeling, including for immune recognition, and highlight considerations regarding the use of AlphaFold confidence metrics in model selection.

Plasticity of BioPhi-driven humanness optimization in ScFv-CD99 binding affinity validated through AlphaFold, HADDOCK, and MD simulations

BioPhi-guided humanization was utilized to enhance the humanness of a humanized single-chain variable fragment targeting CD99, leading to the development of two variants: HuScFvMT99/3BP and HuScFvMT99/3HY. The HuScFvMT99/3BP variant incorporated framework region modifications, leading to modest improvements in humanness, particularly in the VH domain, although the VL domain remained suboptimal. To address this limitation, HuScFvMT99/3HY was designed by combining the VL domain of wild-type with the VH domain of HuScFvMT99/3BP. Molecular dynamics simulations employing AlphaFold2, AlphaFold3, and HADDOCK were performed to evaluate the HuScFv-CD99 peptide complexes. AF2-based simulations demonstrated enhanced binding free energy (ΔGbinding) for both variants compared to HuScFvMT99/3WT. However, ΔGbinding values obtained from AF3 and HD simulations were inconsistent, with HuScFvMT99/3BP exhibiting the weakest binding affinity. While ΔGbinding patterns derived from AlphaFold3 and HADDOCK simulations aligned, amino acid decomposition analysis revealed variations in the interaction coordinates of the predicted complexes. Root-mean-square deviation analysis indicated improved structural stability for HuScFvMT99/3BP (0.975 Å) and HuScFvMT99/3HY (1.075 Å) relative to HuScFvMT99/3WT (1.225 Å). Biolayer interferometry further confirmed that HuScFvMT99/3WT exhibited the highest binding affinity (KD = 1.35 × 10⁻⁷ M) compared to HuScFvMT99/3BP (KD = 2.64 × 10⁻⁷ M) and HuScFvMT99/3HY (KD = 3.95 × 10⁻⁷ M). Supporting evidence was provided by ELISA and flow cytometry experiments. PITHA analysis revealed a high immunogenicity risk for all variants, despite HuScFvMT99/3HY displaying improved humanness, a larger complementarity-determining region (CDR) cavity, and a more hydrophobic CDR-H3 loop. These findings highlight the delicate balance between enhancing humanness and preserving the structural and functional integrity critical for therapeutic antibody development.

Epitope mapping via in vitro deep mutational scanning methods and its applications

Epitope mapping is a technique employed to define the region of an antigen that elicits an immune response, providing crucial insight into the structural architecture of the antigen as well as epitope-paratope interactions. With this breadth of knowledge, immunotherapies, diagnostics, and vaccines are being developed with a rational and data-supported design. Traditional epitope mapping methods are laborious, time-intensive, and often lack the ability to screen proteins in a high-throughput manner or provide high resolution. Deep mutational scanning (DMS), however, is revolutionizing the field as it can screen all possible single amino acid mutations and provide an efficient and high-throughput way to infer the structures of both linear and three-dimensional epitopes with high resolution. Currently, more than 50 publications take this approach to efficiently identify enhancing or escaping mutations, with many then employing this information to rapidly develop broadly neutralizing antibodies, T-cell immunotherapies, vaccine platforms, or diagnostics. We provide a comprehensive review of the approaches to accomplish epitope mapping while also providing a summation of the development of DMS technology and its impactful applications.

Integrative Modeling in the Age of Machine Learning: A Summary of HADDOCK Strategies in CAPRI Rounds 47-55

The HADDOCK team participated in CAPRI rounds 47-55 as server, manual predictor, and scorers. Throughout these CAPRI rounds, we used a plethora of computational strategies to predict the structure of protein complexes. Of the 10 targets comprising 24 interfaces, we achieved acceptable or better models for 3 targets in the human category and 1 in the server category. Our performance in the scoring challenge was slightly better, with our simple scoring protocol being the only one capable of identifying an acceptable model for Target 234. This result highlights the robustness of the simple, fully physics-based HADDOCK scoring function, especially when applied to highly flexible antibody-antigen complexes. Inspired by the significant advances in machine learning for structural biology and the dramatic improvement in our success rates after the public release of Alphafold2, we identify the integration of classical approaches like HADDOCK with AI-driven structure prediction methods as a key strategy for improving the accuracy of model generation and scoring.

Reading the repertoire: Progress in adaptive immune receptor analysis using machine learning

The adaptive immune system holds invaluable information on past and present immune responses in the form of B and T cell receptor sequences, but we are limited in our ability to decode this information. Machine learning approaches are under active investigation for a range of tasks relevant to understanding and manipulating the adaptive immune receptor repertoire, including matching receptors to the antigens they bind, generating antibodies or T cell receptors for use as therapeutics, and diagnosing disease based on patient repertoires. Progress on these tasks has the potential to substantially improve the development of vaccines, therapeutics, and diagnostics, as well as advance our understanding of fundamental immunological principles. We outline key challenges for the field, highlighting the need for software benchmarking, targeted large-scale data generation, and coordinated research efforts.

Predicting adaptive immune receptor specificities by machine learning is a data generation problem

Determining the specificity of adaptive immune receptors-B cell receptors (BCRs), their secreted form antibodies, and T cell receptors (TCRs)-is critical for understanding immune responses and advancing immunotherapy and drug discovery. Immune receptors exhibit extensive diversity in their variable domains, enabling them to interact with a plethora of antigens. Despite the significant progress made by AI tools such as AlphaFold in predicting protein structures, challenges remain in accurately modeling the structure and specificity of immune receptors, primarily due to the limited availability of high-quality crystal structures and the complexity of immune receptor-antigen interactions. In this perspective, we highlight recent advancements in sequence-based and structure-based data generation for immune receptors, which are crucial for training machine learning models that predict receptor specificity. We discuss the current bottlenecks and potential future directions in generating and utilizing high-dimensional datasets for predicting and designing the specificity of antibodies and TCRs.

Structural characterization and AlphaFold modeling of human T cell receptor recognition of NRAS cancer neoantigens

T cell receptors (TCRs) that recognize cancer neoantigens are important for anticancer immune responses and immunotherapy. Understanding the structural basis of TCR recognition of neoantigens provides insights into their exquisite specificity and can enable design of optimized TCRs. We determined crystal structures of a human TCR in complex with NRAS Q61K and Q61R neoantigen peptides and HLA-A1 major histocompatibility complex (MHC), revealing the molecular underpinnings for dual recognition and specificity versus wild-type NRAS peptide. We then used multiple versions of AlphaFold to model the corresponding complex structures, given the challenge of immune recognition for such methods. One implementation of AlphaFold2 (TCRmodel2) with additional sampling was able to generate accurate models of the complexes, while AlphaFold3 also showed strong performance, although success was lower for other complexes. This study provides insights into TCR recognition of a shared cancer neoantigen as well as the utility and practical considerations for using AlphaFold to model TCR-peptide-MHC complexes.

MassiveFold: unveiling AlphaFold's hidden potential with optimized and parallelized massive sampling

Massive sampling in AlphaFold enables access to increased structural diversity. In combination with its efficient confidence ranking, this unlocks elevated modeling capabilities for monomeric structures and foremost for protein assemblies. However, the approach struggles with GPU cost and data storage. Here we introduce MassiveFold, an optimized and customizable version of AlphaFold that runs predictions in parallel, reducing the computing time from several months to hours. MassiveFold is scalable and able to run on anything from a single computer to a large GPU infrastructure, where it can fully benefit from all the computing nodes.

On the humanization of VHHs: Prospective case studies, experimental and computational characterization of structural determinants for functionality

The humanization of camelid-derived variable domain heavy chain antibodies (VHHs) poses challenges including immunogenicity, stability, and potential reduction of affinity. Critical to this process are complementarity-determining regions (CDRs), Vernier and Hallmark residues, shaping the three-dimensional fold and influencing VHH structure and function. Additionally, the presence of non-canonical disulfide bonds further contributes to conformational stability and antigen binding. In this study, we systematically humanized two camelid-derived VHHs targeting the natural cytotoxicity receptor NKp30. Key structural positions in Vernier and Hallmark regions were exchanged with residues from the most similar human germline sequences. The resulting variants were characterized for binding affinities, yield, and purity. Structural binding modes were elucidated through crystal structure determination and AlphaFold2 predictions, providing insights into differences in binding affinity. Comparative structural and molecular dynamics characterizations of selected variants were performed to rationalize their functional properties and elucidate the role of specific sequence motifs in antigen binding. Furthermore, systematic analyses of next-generation sequencing (NGS) and Protein Data Bank (PDB) data was conducted, shedding light on the functional significance of Hallmark motifs and non-canonical disulfide bonds in VHHs in general. Overall, this study provides valuable insights into the structural determinants governing the functional properties of VHHs, offering a roadmap for their rational design, humanization, and optimization for therapeutic applications.

Unmasking AlphaFold to integrate experiments and predictions in multimeric complexes

Since the release of AlphaFold, researchers have actively refined its predictions and attempted to integrate it into existing pipelines for determining protein structures. These efforts have introduced a number of functionalities and optimisations at the latest Critical Assessment of protein Structure Prediction edition (CASP15), resulting in a marked improvement in the prediction of multimeric protein structures. However, AlphaFold's capability of predicting large protein complexes is still limited and integrating experimental data in the prediction pipeline is not straightforward. In this study, we introduce AF_unmasked to overcome these limitations. Our results demonstrate that AF_unmasked can integrate experimental information to build larger or hard to predict protein assemblies with high confidence. The resulting predictions can help interpret and augment experimental data. This approach generates high quality (DockQ score > 0.8) structures even when little to no evolutionary information is available and imperfect experimental structures are used as a starting point. AF_unmasked is developed and optimised to fill incomplete experimental structures (structural inpainting), which may provide insights into protein dynamics. In summary, AF_unmasked provides an easy-to-use method that efficiently integrates experiments to predict large protein complexes more confidently.

Improved deep learning prediction of antigen-antibody interactions

Identifying antibodies that neutralize specific antigens is crucial for developing effective immunotherapies, but this task remains challenging for many target antigens. The rise of deep learning-based computational approaches presents a promising avenue to address this challenge. Here, we assess the performance of a deep learning approach through two benchmark tests aimed at predicting antibodies for the receptor-binding domain of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) spike protein. Three different strategies for constructing input sequence alignments are employed for predicting structural models of antigen-antibody complexes. In our initial testing set, which comprises known experimental structures, these strategies collectively yield a significant top-ranked prediction for 61% of cases and a success rate of 47%. Notably, one strategy that utilizes the sequences of known antigen binders outperforms the other two, achieving a precision of 90% in a subsequent test set of ~1,000 antibodies, balanced between true and control antibodies for the antigen, albeit with a lower recall of 25%. Our results underscore the potential of integrating deep learning methods with single B cell sequencing techniques to enhance the prediction accuracy of antigen-antibody interactions.

Towards the accurate modelling of antibody-antigen complexes from sequence using machine learning and information-driven docking

MOTIVATION

Antibody-antigen complex modelling is an important step in computational workflows for therapeutic antibody design. While experimentally determined structures of both antibody and the cognate antigen are often not available, recent advances in machine learning-driven protein modelling have enabled accurate prediction of both antibody and antigen structures. Here, we analyse the ability of protein-protein docking tools to use machine learning generated input structures for information-driven docking.

RESULTS

In an information-driven scenario, we find that HADDOCK can generate accurate models of antibody-antigen complexes using an ensemble of antibody structures generated by machine learning tools and AlphaFold2 predicted antigen structures. Targeted docking using knowledge of the complementary determining regions on the antibody and some information about the targeted epitope allows the generation of high-quality models of the complex with reduced sampling, resulting in a computationally cheap protocol that outperforms the ZDOCK baseline.

AVAILABILITY AND IMPLEMENTATION

The source code of HADDOCK3 is freely available at github.com/haddocking/haddock3. The code to generate and analyse the data is available at github.com/haddocking/ai-antibodies. The full runs, including docking models from all modules of a workflow have been deposited in our lab collection (data.sbgrid.org/labs/32/1139) at the SBGRID data repository.

Characterization of a Novel Monoclonal Antibody with High Affinity and Specificity against Aflatoxins: A Discovery from Rosetta Antibody-Ligand Computational Simulation

Aflatoxin B1 (AFB1) accumulates in crops, where it poses a threat to human health. To detect AFB1, anti-AFB1 monoclonal antibodies have been developed and are widely used. While the sensitivity and specificity of these antibodies have been extensively studied, information regarding the atomic-level docking of AFB1 (and its derivatives) with these antibodies is limited. Such information is crucial for understanding the key interactions that are required for high affinity and specificity in aflatoxin binding. First, a 3D comparative model of anti-AFB1 antibody (Ab-4B5G6) was predicted from the sequence using RosettaAntibody. We then utilized RosettaLigand to dock AFB1 onto ten homology models, producing a total of 10,000 binding modes. Interestingly, the best-scoring mode predicted strong interactions involving four sites within the heavy chain: ALA33, ASN52, HIS95, and TRP99. Importantly, these strong binding interactions exclusively involve the variable domain of the heavy chain. The best-scoring mode with AFB1 was also obtained through AF multimer combined with RosettaLigand, and two interactions at TRP and HIS were consistent with those found by Rosetta antibody-ligand computational simulation. The role of tryptophan in π interactions in antibodies was confirmed through mutation experiments, and the resulting mutant (W99A) exhibited a >1000-fold reduction in binding affinity for AFB1 and analogs, indicating the effect of tryptophan on the stability of CDR-H3 region. Additionally, we evaluated the binding of two glycolic acid-derived molecular derivatives (with impaired hydrogen bonding potential), and these derivatives (AFB2-GA and AFG2-GA) demonstrated a very weak binding affinity for Ab-4B5G6. The heavy chain was successfully isolated, and its sensitivity and specificity were consistent with those of the intact antibody. The homology models of variable heavy (VH) single-domain antibodies were established by RosettaAntibody, and the docking analysis revealed the same residues, including Ala, His, and Trp. Compared to the potential binding mode of fragment variable (FV) region, the results from a model of VH indicated that there are seven models involved in hydrophobic interaction with TYR32, which is usually referred to as polar amino acid and has both hydrophobic and hydrophilic features depending on the circumstances. Our work encompasses the entire process of Rosetta antibody-ligand computational simulation, highlighting the significance of variable heavy domain structural design in enhancing molecular interactions.

A comparison of antibody-antigen complex sequence-to-structure prediction methods and their systematic biases

The ability to accurately predict antibody-antigen complex structures from their sequences could greatly advance our understanding of the immune system and would aid in the development of novel antibody therapeutics. There have been considerable recent advancements in predicting protein-protein interactions (PPIs) fueled by progress in machine learning (ML). To understand the current state of the field, we compare six representative methods for predicting antibody-antigen complexes from sequence, including two deep learning approaches trained to predict PPIs in general (AlphaFold-Multimer and RoseTTAFold), two composite methods that initially predict antibody and antigen structures separately and dock them (using antibody-mode ClusPro), local refinement in Rosetta (SnugDock) of globally docked poses from ClusPro, and a pipeline combining homology modeling with rigid-body docking informed by ML-based epitope and paratope prediction (AbAdapt). We find that AlphaFold-Multimer outperformed other methods, although the absolute performance leaves considerable room for improvement. AlphaFold-Multimer models of lower quality display significant structural biases at the level of tertiary motifs (TERMs) toward having fewer structural matches in non-antibody-containing structures from the Protein Data Bank (PDB). Specifically, better models exhibit more common PDB-like TERMs at the antibody-antigen interface than worse ones. Importantly, the clear relationship between performance and the commonness of interfacial TERMs suggests that the scarcity of interfacial geometry data in the structural database may currently limit the application of ML to the prediction of antibody-antigen interactions.

TEMPRO: nanobody melting temperature estimation model using protein embeddings

Single-domain antibodies (sdAbs) or nanobodies have received widespread attention due to their small size (~ 15 kDa) and diverse applications in bio-derived therapeutics. As many modern biotechnology breakthroughs are applied to antibody engineering and design, nanobody thermostability or melting temperature (Tm) is crucial for their successful utilization. In this study, we present TEMPRO which is a predictive modeling approach for estimating the Tm of nanobodies using computational methods. Our methodology integrates various nanobody biophysical features to include Evolutionary Scale Modeling (ESM) embeddings, NetSurfP3 structural predictions, pLDDT scores per sdAb region from AlphaFold2, and each sequence's physicochemical characteristics. This approach is validated with our combined dataset containing 567 unique sequences with corresponding experimental Tm values from a manually curated internal data and a recently published nanobody database, NbThermo. Our results indicate the efficacy of protein embeddings in reliably predicting the Tm of sdAbs with mean absolute error (MAE) of 4.03 °C and root mean squared error (RMSE) of 5.66 °C, thus offering a valuable tool for the optimization of nanobodies for various biomedical and therapeutic applications. Moreover, we have validated the models' performance using experimentally determined Tms from nanobodies not found in NbThermo. This predictive model not only enhances nanobody thermostability prediction, but also provides a useful perspective of using embeddings as a tool for facilitating a broader applicability of downstream protein analyses.

Integrating Large-Scale Protein Structure Prediction into Human Genetics Research

The last five years have seen impressive progress in deep learning models applied to protein research. Most notably, sequence-based structure predictions have seen transformative gains in the form of AlphaFold2 and related approaches. Millions of missense protein variants in the human population lack annotations, and these computational methods are a valuable means to prioritize variants for further analysis. Here, we review the recent progress in deep learning models applied to the prediction of protein structure and protein variants, with particular emphasis on their implications for human genetics and health. Improved prediction of protein structures facilitates annotations of the impact of variants on protein stability, protein-protein interaction interfaces, and small-molecule binding pockets. Moreover, it contributes to the study of host-pathogen interactions and the characterization of protein function. As genome sequencing in large cohorts becomes increasingly prevalent, we believe that better integration of state-of-the-art protein informatics technologies into human genetics research is of paramount importance.

The power and pitfalls of AlphaFold2 for structure prediction beyond rigid globular proteins

Artificial intelligence-driven advances in protein structure prediction in recent years have raised the question: has the protein structure-prediction problem been solved? Here, with a focus on nonglobular proteins, we highlight the many strengths and potential weaknesses of DeepMind's AlphaFold2 in the context of its biological and therapeutic applications. We summarize the subtleties associated with evaluation of AlphaFold2 model quality and reliability using the predicted local distance difference test (pLDDT) and predicted aligned error (PAE) values. We highlight various classes of proteins that AlphaFold2 can be applied to and the caveats involved. Concrete examples of how AlphaFold2 models can be integrated with experimental data in the form of small-angle X-ray scattering (SAXS), solution NMR, cryo-electron microscopy (cryo-EM) and X-ray diffraction are discussed. Finally, we highlight the need to move beyond structure prediction of rigid, static structural snapshots toward conformational ensembles and alternate biologically relevant states. The overarching theme is that careful consideration is due when using AlphaFold2-generated models to generate testable hypotheses and structural models, rather than treating predicted models as de facto ground truth structures.

Integrative modeling meets deep learning: Recent advances in modeling protein assemblies

Recent progress in protein structure prediction based on deep learning revolutionized the field of Structural Biology. Beyond single proteins, it also enabled high-throughput prediction of structures of protein-protein interactions. Despite the success in predicting complex structures, large macromolecular assemblies still require specialized approaches. Here we describe recent advances in modeling macromolecular assemblies using integrative and hierarchical approaches. We highlight applications that predict protein-protein interactions and challenges in modeling complexes based on the interaction networks, including the prediction of complex stoichiometry and heterogeneity.

AlphaFold2 in biomedical research: facilitating the development of diagnostic strategies for disease

Proteins, as the primary executors of physiological activity, serve as a key factor in disease diagnosis and treatment. Research into their structures, functions, and interactions is essential to better understand disease mechanisms and potential therapies. DeepMind's AlphaFold2, a deep-learning protein structure prediction model, has proven to be remarkably accurate, and it is widely employed in various aspects of diagnostic research, such as the study of disease biomarkers, microorganism pathogenicity, antigen-antibody structures, and missense mutations. Thus, AlphaFold2 serves as an exceptional tool to bridge fundamental protein research with breakthroughs in disease diagnosis, developments in diagnostic strategies, and the design of novel therapeutic approaches and enhancements in precision medicine. This review outlines the architecture, highlights, and limitations of AlphaFold2, placing particular emphasis on its applications within diagnostic research grounded in disciplines such as immunology, biochemistry, molecular biology, and microbiology.

Geometric epitope and paratope prediction

MOTIVATION

Identifying the binding sites of antibodies is essential for developing vaccines and synthetic antibodies. In this article, we investigate the optimal representation for predicting the binding sites in the two molecules and emphasize the importance of geometric information.

RESULTS

Specifically, we compare different geometric deep learning methods applied to proteins' inner (I-GEP) and outer (O-GEP) structures. We incorporate 3D coordinates and spectral geometric descriptors as input features to fully leverage the geometric information. Our research suggests that different geometrical representation information is useful for different tasks. Surface-based models are more efficient in predicting the binding of the epitope, while graph models are better in paratope prediction, both achieving significant performance improvements. Moreover, we analyze the impact of structural changes in antibodies and antigens resulting from conformational rearrangements or reconstruction errors. Through this investigation, we showcase the robustness of geometric deep learning methods and spectral geometric descriptors to such perturbations.

AVAILABILITY AND IMPLEMENTATION

The python code for the models, together with the data and the processing pipeline, is open-source and available at https://github.com/Marco-Peg/GEP.

A Comparison of Antibody-Antigen Complex Sequence-to-Structure Prediction Methods and their Systematic Biases

The ability to accurately predict antibody-antigen complex structures from their sequences could greatly advance our understanding of the immune system and would aid in the development of novel antibody therapeutics. There have been considerable recent advancements in predicting protein-protein interactions (PPIs) fueled by progress in machine learning (ML). To understand the current state of the field, we compare six representative methods for predicting antibody-antigen complexes from sequence, including two deep learning approaches trained to predict PPIs in general (AlphaFold-Multimer, RoseTTAFold), two composite methods that initially predict antibody and antigen structures separately and dock them (using antibody-mode ClusPro), local refinement in Rosetta (SnugDock) of globally docked poses from ClusPro, and a pipeline combining homology modeling with rigid-body docking informed by ML-based epitope and paratope prediction (AbAdapt). We find that AlphaFold-Multimer outperformed other methods, although the absolute performance leaves considerable room for improvement. AlphaFold-Multimer models of lower-quality display significant structural biases at the level of tertiary motifs (TERMs) towards having fewer structural matches in non-antibody containing structures from the Protein Data Bank (PDB). Specifically, better models exhibit more common PDB-like TERMs at the antibody-antigen interface than worse ones. Importantly, the clear relationship between performance and the commonness of interfacial TERMs suggests that scarcity of interfacial geometry data in the structural database may currently limit application of machine learning to the prediction of antibody-antigen interactions.

Overview of AlphaFold2 and breakthroughs in overcoming its limitations

Predicting three-dimensional (3D) protein structures has been challenging for decades. The emergence of AlphaFold2 (AF2), a deep learning-based machine learning method developed by DeepMind, became a game changer in the protein folding community. AF2 can predict a protein's three-dimensional structure with high confidence based on its amino acid sequence. Accurate prediction of protein structures can dramatically accelerate our understanding of biological mechanisms and provide a solid foundation for reliable drug design. Although AF2 breaks through the barriers in predicting protein structures, many rooms remain to be further studied. This review provides a brief historical overview of the development of protein structure prediction, covering template-based, template-free, and machine learning-based methods. In addition to reviewing the potential benefits (Pros) and considerations (Cons) of using AF2, this review summarizes the diverse applications, including protein structure predictions, dynamic changes, point mutation, integration of language model and experimental data, protein complex, and protein-peptide interaction. It underscores recent advancements in efficiency, reliability, and broad application of AF2. This comprehensive review offers valuable insights into the applications of AF2 and AF2-inspired AI methods in structural biology and its potential for clinically significant drug target discovery.

Structural characterization and AlphaFold modeling of human T cell receptor recognition of NRAS cancer neoantigens

T cell receptors (TCRs) that recognize cancer neoantigens are important for anti-cancer immune responses and immunotherapy. Understanding the structural basis of TCR recognition of neoantigens provides insights into their exquisite specificity and can enable design of optimized TCRs. We determined crystal structures of a human TCR in complex with NRAS Q61K and Q61R neoantigen peptides and HLA-A1 MHC, revealing the molecular underpinnings for dual recognition and specificity versus wild-type NRAS peptide. We then used multiple versions of AlphaFold to model the corresponding complex structures, given the challenge of immune recognition for such methods. Interestingly, one implementation of AlphaFold2 (TCRmodel2) was able to generate accurate models of the complexes, while AlphaFold3 also showed strong performance, although success was lower for other complexes. This study provides insights into TCR recognition of a shared cancer neoantigen, as well as the utility and practical considerations for using AlphaFold to model TCR-peptide-MHC complexes.

Experimental Structures of Antibody/MHC-I Complexes Reveal Details of Epitopes Overlooked by Computational Prediction

mAbs to MHC class I (MHC-I) molecules have proved to be crucial reagents for tissue typing and fundamental studies of immune recognition. To augment our understanding of epitopic sites seen by a set of anti-MHC-I mAb, we determined X-ray crystal structures of four complexes of anti-MHC-I Fabs bound to peptide/MHC-I/β2-microglobulin (pMHC-I). An anti-H2-Dd mAb, two anti-MHC-I α3 domain mAbs, and an anti-β2-microglobulin mAb bind pMHC-I at sites consistent with earlier mutational and functional experiments, and the structures explain allelomorph specificity. Comparison of the experimentally determined structures with computationally derived models using AlphaFold Multimer showed that although predictions of the individual pMHC-I heterodimers were quite acceptable, the computational models failed to properly identify the docking sites of the mAb on pMHC-I. The experimental and predicted structures provide insight into strengths and weaknesses of purely computational approaches and suggest areas that merit additional attention.

Flexible protein-protein docking with a multitrack iterative transformer

Conventional protein-protein docking algorithms usually rely on heavy candidate sampling and reranking, but these steps are time-consuming and hinder applications that require high-throughput complex structure prediction, for example, structure-based virtual screening. Existing deep learning methods for protein-protein docking, despite being much faster, suffer from low docking success rates. In addition, they simplify the problem to assume no conformational changes within any protein upon binding (rigid docking). This assumption precludes applications when binding-induced conformational changes play a role, such as allosteric inhibition or docking from uncertain unbound model structures. To address these limitations, we present GeoDock, a multitrack iterative transformer network to predict a docked structure from separate docking partners. Unlike deep learning models for protein structure prediction that input multiple sequence alignments, GeoDock inputs just the sequences and structures of the docking partners, which suits the tasks when the individual structures are given. GeoDock is flexible at the protein residue level, allowing the prediction of conformational changes upon binding. On the Database of Interacting Protein Structures (DIPS) test set, GeoDock achieves a 43% top-1 success rate, outperforming all other tested methods. However, in the standard DIPS train/test splits, we discovered contamination of close homologs in the training set. After decontaminating the training set, the success rate is 31%. On the DB5.5 test set and a benchmark dataset of antibody-antigen complexes, GeoDock outperforms the deep learning models trained using the same dataset but falls behind most of the conventional methods and AlphaFold-Multimer. GeoDock attains an average inference speed of under 1 s on a single GPU, enabling its application in large-scale structure screening. Although binding-induced conformational changes are still a challenge owing to limited training and evaluation data, our architecture sets up the foundation to capture this backbone flexibility. Code and a demonstration Jupyter notebook are available at https://github.com/Graylab/GeoDock.

Applications and challenges in designing VHH-based bispecific antibodies: leveraging machine learning solutions

The development of bispecific antibodies that bind at least two different targets relies on bringing together multiple binding domains with different binding properties and biophysical characteristics to produce a drug-like therapeutic. These building blocks play an important role in the overall quality of the molecule and can influence many important aspects from potency and specificity to stability and half-life. Single-domain antibodies, particularly camelid-derived variable heavy domain of heavy chain (VHH) antibodies, are becoming an increasingly popular choice for bispecific construction due to their single-domain modularity, favorable biophysical properties, and potential to work in multiple antibody formats. Here, we review the use of VHH domains as building blocks in the construction of multispecific antibodies and the challenges in creating optimized molecules. In addition to exploring traditional approaches to VHH development, we review the integration of machine learning techniques at various stages of the process. Specifically, the utilization of machine learning for structural prediction, lead identification, lead optimization, and humanization of VHH antibodies.