Abstract:Antibody-based therapies are at the forefront of modern medicine, addressing diverse challenges across oncology, autoimmune diseases, infectious diseases, and beyond. The ability to design antibodies with enhanced functionality and specificity is critical for advancing next-generation therapeutics. Recent advances in artificial intelligence (AI) have propelled the field of antibody engineering, particularly through inverse folding models for Complementarity-Determining Region (CDR) sequence design. These models aim to generate novel antibody sequences that fold into desired structures with high antigen-binding affinity. However, current evaluation metrics, such as amino acid recovery rates, are limited in their ability to assess the structural and functional accuracy of designed sequences. This study benchmarks state-of-the-art inverse folding models - ProteinMPNN, ESM-IF, LM-Design, and AntiFold - using comprehensive datasets and alternative evaluation metrics like sequence similarity. By systematically analyzing recovery rates, mutation prediction capabilities, and amino acid composition biases, we identify strengths and limitations across models. AntiFold exhibits superior performance in Fab antibody design, particularly in variable regions like CDRH3, whereas LM-Design demonstrates adaptability across diverse antibody types, including VHH antibodies. In contrast, models trained on general protein datasets (e.g., ProteinMPNN and ESM-IF) struggle with antibody-specific nuances. Key insights include the models' varying reliance on antigen structure and their distinct capabilities in capturing critical residues for antigen binding. Our findings highlight the need for enhanced training datasets, integration of functional data, and refined evaluation metrics to advance antibody design tools. By addressing these challenges, future models can unlock the full potential of AI-driven antibody engineering, paving the way for innovative therapeutic applications.

Structural Modeling of Nanobodies: A Benchmark of State-of-the-Art Artificial Intelligence Programs

Benchmarking Inverse Folding Models for Antibody CDR Sequence Design

Nanobody Engineering: Computational Modelling and Design for Biomedical and Therapeutic Applications

AI driven approaches in Nanobody Epitope Prediction: Are We There Yet?

NbX: Machine Learning-Guided Re-Ranking of Nanobody–Antigen Binding Poses

AI-Augmented Physics-Based Docking for Antibody-Antigen Complex Prediction

ABodyBuilder3: Improved and scalable antibody structure predictions

IgGM: A Generative Model for Functional Antibody and Nanobody Design

Accurate Prediction of Antibody Function and Structure Using Bio-Inspired Antibody Language Model

AI-based IsAb2.0 for antibody design

Advancements in nanobody generation: Integrating conventional, in silico, and machine learning approaches

Nanobodies: from Structure to Applications in Non-Injectable and Bispecific Biotherapeutic Development.

Structural modeling of antibody variable regions using deep learning—progress and perspectives on drug discovery

What has AlphaFold3 learned about antibody and nanobody docking, and what remains unsolved?

Benchmarking AlphaFold for protein complex modeling reveals accuracy determinants

Accurate prediction of CDR-H3 loop structures of antibodies with deep learning

xTrimoABFold: Improving Antibody Structure Prediction without Multiple Sequence Alignments