Protein-protein interaction prediction methods: from docking-based to AI-based approaches

Computational methods for modeling protein-protein interactions in the AI era: Current status and future directions

The modeling of protein-protein interactions (PPIs) has been revolutionized by artificial intelligence, with deep learning and end-to-end frameworks such as AlphaFold and its derivatives now dominating the field. This review surveys the current computational landscape for predicting protein complex structures, outlining the role of traditional docking approaches as well as focusing on recent advances in AI-driven methods. We discuss key challenges, including protein flexibility, reliance on co-evolutionary signals, modeling of large assemblies, and interactions involving intrinsically disordered regions (IDRs). Recent innovations aimed at improving sampling diversity, integrating experimental data, and enhancing robustness are also highlighted. Although classical methods remain relevant in specific contexts, the continued evolution of AI-based tools offers transformative potential for structural biology. These advances are poised to deepen our understanding of biomolecular interactions and accelerate the design of therapeutic interventions.

Sensory plasticity caused by up-down regulation encodes the information of short-term learning and memory

Learning and memory are essential for animals' well-being and survival. The underlying mechanisms are a major task of neuroscience studies. In this study, we identified a circuit consisting of ASER, RIC, RIS, and AIY, is required for short-term salt chemotaxis learning (SCL) in C. elegans. ASER NaCl-sensation possesses are remodeled by salt/food-deprivation pared conditioning. RIC integrates the sensory information of NaCl and food availability. It excites ASER and inhibits AIY by tyramine/TYRA-2 and octopamine/OCTR-1 signaling pathways, respectively. By the salt conditioning, RIC NaCl calcium response to NaCl is depressed, thus, the RIC excitation of ASER and inhibition of AIY are suppressed. ASER excites RIS by FLP-14/FRPR-10 signaling. RIS inhibits ASER via PDF-2/PDFR-1 signaling in negative feedback. ASER sensory plasticity caused by RIC plasticity and RIS negative feedback are required for both learning and memory recall. Thus, the sensation plasticity encodes the information of the short-term SCL that facilitates animal adaptation to dynamic environments.

Comprehensive Atomic-Scale 3D Viral-Host Protein Interactomes Enable Dissection of Key Mechanisms and Evolutionary Processes Underlying Viral Pathogenesis

Viral-human protein interactions are critical for viral replication and modulation of the host immune response. Structural modeling of these interactions is vital for developing effective antiviral therapies and vaccines. However, 99% of experimentally determined binary host-viral interactions currently lack structural information. We aimed to address this gap by leveraging computational protein structure prediction methods. Using extensive benchmarking, we found AlphaFold to be the most accurate structure prediction model for host-pathogen protein interactions. We then predicted the structures of 11,666 binary protein interactions across 33 viral families and created the most comprehensive atomic-scale 3D viral-host protein interactomes till date ( https://3d-viralhuman.yulab.org ). By integrating these interactomes with genetic variation data, we identified population-specific signatures of selection on variants coding for interfaces of viral-human interactions. We also found that viral interaction interfaces were less conserved than non-interface regions, a striking trend that is opposite to what is observed for host interfaces, suggesting different evolutionary pressures. Systematic analyses of interface sharing between host and viral proteins binding to the same host protein revealed mutation rate-dependent differences in interface mimicry. Similar mutation rate-dependent differences were seen in the interface sharing between viral proteins binding to a host protein. We also found that the patterns of E6 protein binding to KPNA2 differed between high- and low-risk oncogenic human papillomaviruses (HPVs), and clustering based on these binding patterns allowed the classification of HPVs with unknown oncogenic risk. Our interface mimicry analyses also unveiled a novel mechanism by which herpes simplex virus-1 UL37 suppresses the antiviral immune response through disruption of the TRAF6-MAVS signalosome interaction. Overall, our comprehensive 3D viral interactomes provide a resource at unprecedented scale and resolution that will enable researchers to explore how variation and signatures of selection influence viral interactions and disease progression. This tool also facilitates the identification of conserved and unique interaction patterns across viruses, empowering researchers to generate testable hypotheses and ultimately accelerate the discovery of novel therapeutic targets and intervention strategies.

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.

Barley2035: A decadal vision for barley research and breeding

Barley (Hordeum vulgare ssp. vulgare) is one of the oldest founder crops in human civilization and has been widely dispersed across the globe to support human society as a livestock feed and a raw material for the brewing industries. Since the early half of the 20th century, it has been used for innovative research on cytogenetics, biochemistry, and genetics, facilitated by its mode of reproduction through self-pollination and its true diploid status, which have contributed to the accumulation of numerous germplasm and mutant resources. In the era of molecular genomics and biology, a multitude of barley genes and their related regulatory mechanisms have been identified and functionally validated, providing a paradigm for equivalent studies in other Triticeae crops. This review highlights important advances on barley research over the past decade, focusing mainly on genomics and genomics-assisted germplasm exploration, genetic dissection of developmental and adaptation-related traits, and the complex dynamics of yield and quality formation. In the coming decade, the prospect of integrating these innovations in barley research and breeding shows great promise. Barley is proposed as a reference Triticeae crop for the discovery and functional validation of new genes and the dissection of their molecular mechanisms. The application of precise genome editing as well as genomic prediction and selection, further enhanced by artificial intelligence-based tools and applications, is expected to promote barley improvement to efficiently meet the evolving global demands for this important crop.

Martínez-Castilla L, García-Sandoval R, Ballester P, Ferrándiz C, Sanchez MDLP, García-Ponce B, Garay-Arroyo A, R. Álvarez-Buylla E. HDACs MADS-domain protein interaction: a case study of HDA15 and XAL1 in Arabidopsis thaliana

Cellular behavior, cell differentiation and ontogenetic development in eukaryotes result from complex interactions between epigenetic and classic molecular genetic mechanisms, with many of these interactions still to be elucidated. Histone deacetylase enzymes (HDACs) promote the interaction of histones with DNA by compacting the nucleosome, thus causing transcriptional repression. MADS-domain transcription factors are highly conserved in eukaryotes and participate in controlling diverse developmental processes in animals and plants, as well as regulating stress responses in plants. In this work, we focused on finding out putative interactions of Arabidopsis thaliana HDACs and MADS-domain proteins using an evolutionary perspective combined with bioinformatics analyses and testing the more promising predicted interactions through classic molecular biology tools. Through bioinformatic analyses, we found similarities between HDACs proteins from different organisms, which allowed us to predict a putative protein-protein interaction between the Arabidopsis thaliana deacetylase HDA15 and the MADS-domain protein XAANTAL1 (XAL1). The results of two-hybrid and Bimolecular Fluorescence Complementation analysis demonstrated in vitro and in vivo HDA15-XAL1 interaction in the nucleus. Likely, this interaction might regulate developmental processes in plants as is the case for this type of interaction in animals.

ProAffinity-GNN: A Novel Approach to Structure-Based Protein-Protein Binding Affinity Prediction via a Curated Data Set and Graph Neural Networks

Protein-protein interactions (PPIs) are crucial for understanding biological processes and disease mechanisms, contributing significantly to advances in protein engineering and drug discovery. The accurate determination of binding affinities, essential for decoding PPIs, faces challenges due to the substantial time and financial costs involved in experimental and theoretical methods. This situation underscores the urgent need for more effective and precise methodologies for predicting binding affinity. Despite the abundance of research on PPI modeling, the field of quantitative binding affinity prediction remains underexplored, mainly due to a lack of comprehensive data. This study seeks to address these needs by manually curating pairwise interaction labels on available 3D structures of protein complexes, with experimentally determined binding affinities, creating the largest data set for structure-based pairwise protein interaction with binding affinity to date. Subsequently, we introduce ProAffinity-GNN, a novel deep learning framework using protein language model and graph neural network (GNN) to improve the accuracy of prediction of structure-based protein-protein binding affinities. The evaluation results across several benchmark test sets and an additional case study demonstrate that ProAffinity-GNN not only outperforms existing models in terms of accuracy but also shows strong generalization capabilities.

Repurposing of sericin combined with dactolisib or vitamin D to combat non-small lung cancer cells through computational and biological investigations

This study aims to repurpose sericin in combating non-small lung cancer cells (A549 and H460) by combining it with dactolisib or vitamin D to reduce the dose of dactolisib and boost the anticancer effectiveness of dactolisib and vitamin D. Therefore, the binding affinities of individual and combined drugs were examined using in silico and protein-protein interaction studies, targeting NF-κB, Cyclin D1, p-AKT, and VEGF1 proteins. The findings manifested remarkable affinities for combinatorial drugs compared to individual compounds. To substantiate these findings, the combined IC50 for each combination (sericin + dactolisib and sericin + vitamin D) were determined, reporting 31.9 and 41.8 µg/ml, respectively, against A549 cells and 47.9 and 55.3 µg/ml, respectively, against H460 cells. Furthermore, combination indices were assessed to lower the doses of each drug. Interestingly, in vitro results exhibited marked diminutions in NF-κB, Cyclin D1, p-AKT, and VEGF1 after treatment with sericin + dactolisib and sericin + vitamin D compared to control lung cancer cells and those treated with a single drug. Moreover, A549 and H460 cells treated with both combinations demonstrated augmented caspase-3 levels, implying substantial apoptotic activity. Altogether, these results accentuated the prospective implementation of sericin in combination with dactolisib and vitamin D at low doses to preclude lung cancer cell proliferation.

EuDockScore: Euclidean graph neural networks for scoring protein-protein interfaces

MOTIVATION

Protein-protein interactions are essential for a variety of biological phenomena including mediating biochemical reactions, cell signaling, and the immune response. Proteins seek to form interfaces which reduce overall system energy. Although determination of single polypeptide chain protein structures has been revolutionized by deep learning techniques, complex prediction has still not been perfected. Additionally, experimentally determining structures is incredibly resource and time expensive. An alternative is the technique of computational docking, which takes the solved individual structures of proteins to produce candidate interfaces (decoys). Decoys are then scored using a mathematical function that assess the quality of the system, known as scoring functions. Beyond docking, scoring functions are a critical component of assessing structures produced by many protein generative models. Scoring models are also used as a final filtering in many generative deep learning models including those that generate antibody binders, and those which perform docking.

RESULTS

In this work, we present improved scoring functions for protein-protein interactions which utilizes cutting-edge Euclidean graph neural network architectures, to assess protein-protein interfaces. These Euclidean docking score models are known as EuDockScore, and EuDockScore-Ab with the latter being antibody-antigen dock specific. Finally, we provided EuDockScore-AFM a model trained on antibody-antigen outputs from AlphaFold-Multimer (AFM) which proves useful in reranking large numbers of AFM outputs.

AVAILABILITY AND IMPLEMENTATION

The code for these models is available at https://gitlab.com/mcfeemat/eudockscore.

Connecting the dots: Computational network analysis for disease insight and drug repurposing

Network biology is a powerful framework for studying the structure, function, and dynamics of biological systems, offering insights into the balance between health and disease states. The field is seeing rapid progress in all of its aspects: data availability, network synthesis, network analytics, and impactful applications in medicine and drug development. We review the most recent and significant results in network biomedicine, with a focus on the latest data, analytics, software resources, and applications in medicine. We also discuss what in our view are the likely directions of impactful development over the next few years.

Revolutionizing Molecular Design for Innovative Therapeutic Applications through Artificial Intelligence

The field of computational protein engineering has been transformed by recent advancements in machine learning, artificial intelligence, and molecular modeling, enabling the design of proteins with unprecedented precision and functionality. Computational methods now play a crucial role in enhancing the stability, activity, and specificity of proteins for diverse applications in biotechnology and medicine. Techniques such as deep learning, reinforcement learning, and transfer learning have dramatically improved protein structure prediction, optimization of binding affinities, and enzyme design. These innovations have streamlined the process of protein engineering by allowing the rapid generation of targeted libraries, reducing experimental sampling, and enabling the rational design of proteins with tailored properties. Furthermore, the integration of computational approaches with high-throughput experimental techniques has facilitated the development of multifunctional proteins and novel therapeutics. However, challenges remain in bridging the gap between computational predictions and experimental validation and in addressing ethical concerns related to AI-driven protein design. This review provides a comprehensive overview of the current state and future directions of computational methods in protein engineering, emphasizing their transformative potential in creating next-generation biologics and advancing synthetic biology.

Formulation of next-generation polyvalent vaccine candidates against three important poxviruses by targeting DNA-dependent RNA polymerase using an integrated immunoinformatics and molecular modeling approach

BACKGROUND

Poxviruses comprise a group of large double-stranded DNA viruses and are known to cause diseases in humans, livestock animals, and other animal species. The Mpox virus (MPXV; formerly Monkeypox), variola virus (VARV), and volepox virus (VPXV) are among the prevalent poxviruses of the Orthopoxviridae genera. The ongoing Mpox infectious disease pandemic caused by the Mpox virus has had a major impact on public health across the globe. To date, only limited repurposed antivirals and vaccines are available for the effective treatment of Mpox and other poxviruses that cause contagious diseases.

METHODS

The present study was conducted with the primary goal of formulating multi-epitope vaccines against three evolutionary closed poxviruses i.e., MPXV, VARV, and VPXV using an integrated immunoinformatics and molecular modeling approach. DNA-dependent RNA polymerase (DdRp), a potential vaccine target of poxviruses, has been used to determine immunodominant B and T-cell epitopes followed by interactions analysis with Toll-like receptor 2 at the atomic level.

RESULTS

Three multi-epitope vaccine constructs, namely DdRp_MPXV (V1), DdRp_VARV (V2), and DdRp_VPXV (V3) were designed. These vaccine constructs were found to be antigenic, non-allergenic, non-toxic, and soluble with desired physicochemical properties. Protein-protein docking and interaction profiling analysis depicts a strong binding pattern between the targeted immune receptor TLR2 and the structural models of the designed vaccine constructs, and manifested a number of biochemical bonds (hydrogen bonds, salt bridges, and non-bonded contacts). State-of-the-art all-atoms molecular dynamics simulations revealed highly stable interactions of vaccine constructs with TLR2 at the atomic level throughout the simulations on 300 nanoseconds. Additionally, the outcome of the immune simulation analysis suggested that designed vaccines have the potential to induce protective immunity against targeted poxviruses.

CONCLUSIONS

Taken together, formulated next-generation polyvalent vaccines were found to have good efficacy against closely related poxviruses (MPXV, VARV, and VPXV) as demonstrated by our extensive immunoinformatics and molecular modeling evaluations; however, further experimental investigations are still needed.

A Perspective on the Prospective Use of AI in Protein Structure Prediction

AlphaFold2 (AF2) and RoseTTaFold (RF) have revolutionized structural biology, serving as highly reliable and effective methods for predicting protein structures. This article explores their impact and limitations, focusing on their integration into experimental pipelines and their application in diverse protein classes, including membrane proteins, intrinsically disordered proteins (IDPs), and oligomers. In experimental pipelines, AF2 models help X-ray crystallography in resolving the phase problem, while complementarity with mass spectrometry and NMR data enhances structure determination and protein flexibility prediction. Predicting the structure of membrane proteins remains challenging for both AF2 and RF due to difficulties in capturing conformational ensembles and interactions with the membrane. Improvements in incorporating membrane-specific features and predicting the structural effect of mutations are crucial. For intrinsically disordered proteins, AF2's confidence score (pLDDT) serves as a competitive disorder predictor, but integrative approaches including molecular dynamics (MD) simulations or hydrophobic cluster analyses are advocated for accurate dynamics representation. AF2 and RF show promising results for oligomeric models, outperforming traditional docking methods, with AlphaFold-Multimer showing improved performance. However, some caveats remain in particular for membrane proteins. Real-life examples demonstrate AF2's predictive capabilities in unknown protein structures, but models should be evaluated for their agreement with experimental data. Furthermore, AF2 models can be used complementarily with MD simulations. In this Perspective, we propose a "wish list" for improving deep-learning-based protein folding prediction models, including using experimental data as constraints and modifying models with binding partners or post-translational modifications. Additionally, a meta-tool for ranking and suggesting composite models is suggested, driving future advancements in this rapidly evolving field.

Simplifying complex antibody engineering using machine learning

Machine learning is transforming antibody engineering by enabling the generation of drug-like monoclonal antibodies with unprecedented efficiency. Unsupervised algorithms trained on massive and diverse protein sequence datasets facilitate the prediction of panels of antibody variants with native-like intrinsic properties (e.g., high stability), greatly reducing the amount of subsequent experimentation needed to identify specific candidates that also possess desired extrinsic properties (e.g., high affinity). Additionally, supervised algorithms, which are trained on deep sequencing datasets obtained after enrichment of in vitro antibody libraries for one or more specific extrinsic properties, enable the prediction of antibody variants with desired combinations of extrinsic properties without the need for additional screening. Here we review recent advances using both machine learning approaches and how they are impacting the field of antibody engineering as well as key outstanding challenges and opportunities for these paradigm-changing methods.

Growing ecosystem of deep learning methods for modeling protein-protein interactions

Numerous cellular functions rely on protein-protein interactions. Efforts to comprehensively characterize them remain challenged however by the diversity of molecular recognition mechanisms employed within the proteome. Deep learning has emerged as a promising approach for tackling this problem by exploiting both experimental data and basic biophysical knowledge about protein interactions. Here, we review the growing ecosystem of deep learning methods for modeling protein interactions, highlighting the diversity of these biophysically informed models and their respective trade-offs. We discuss recent successes in using representation learning to capture complex features pertinent to predicting protein interactions and interaction sites, geometric deep learning to reason over protein structures and predict complex structures, and generative modeling to design de novo protein assemblies. We also outline some of the outstanding challenges and promising new directions. Opportunities abound to discover novel interactions, elucidate their physical mechanisms, and engineer binders to modulate their functions using deep learning and, ultimately, unravel how protein interactions orchestrate complex cellular behaviors.

Computational biophysics and structural biology of proteins-a Special Issue in honor of Prof. Haruki Nakamura's 70th birthday

Receiving his initial training jointly in theoretical and applied physics at the University of Tokyo, Professor Haruki Nakamura has had a long and eventful scientific career, along the way helping to shape the way that biophysics is carried out in Japan. Concentrating his research efforts on the simulation of protein structure and function, he has, over his career arc, acted as director of the Institute for Protein Research (Osaka, Japan), director of the Protein Data Bank of Japan (PDBj), president of the Biophysical Society of Japan (BSJ), president of the Protein Science Society of Japan (PSSJ), and group leader and professor of Bioinformatics and Computational Structural Biology at Osaka University. In 2022, Prof. Haruki Nakamura turned 70 years old, and to mark this occasion, his scientific colleagues from around the world have combined their efforts to produce this Festschrift Issue of the IUPAB Biophysical Reviews journal around the theme of the computational biophysics and structural biology of proteins.