Related Experiment Video
Updated: Sep 17, 2025
08:09
Peptide Scanning-assisted Identification of a Monoclonal Antibody-recognized Linear B-cell Epitope
Published on: March 24, 2017
9.6K
AntiBMPNN: Structure-Guided Graph Neural Networks for Precision Antibody Engineering.
Ze-Yu Sun1,2, Jiayi Yuan2, Divya Jaiswal2
1College of Pharmacology Sciences, Zhejiang University of Technology, Hangzhou, 310014, P. R. China.
Advanced Science (Weinheim, Baden-Wurttemberg, Germany)
|June 27, 2025
Summary
A new deep-learning framework, AntiBMPNN, significantly improves antibody sequence design accuracy and binding affinity compared to existing methods. This advancement holds great potential for developing more effective therapeutic antibodies.
Area of Science:
- Biotechnology
- Computational Biology
- Immunology
Background:
- Traditional antibody sequence design methods are inefficient.
- Antibodies are critical for various medical applications.
Purpose of the Study:
- To introduce AntiBMPNN, a novel deep-learning framework for highly accurate antibody sequence design.
- To evaluate AntiBMPNN's performance against existing state-of-the-art methods.
Main Methods:
- Utilized an antibody-specific 3D dataset and a fine-tuned message-passing neural network (MPNN).
- Incorporated a frequency-based scoring function and AlphaFold 3 for sequence prioritization.
- Employed experimental validation for single-point antibody design and binding affinity assessments.
Main Results:
- AntiBMPNN achieved over 80% sequence recovery and a perplexity of 1.5, outperforming ProteinMPNN.
- Demonstrated a 75% success rate in single-point antibody design.
- Designed sequences with significantly enhanced binding affinities, notably for CDR1, CDR2, and CDR3 regions of specific nanobodies, outperforming multiple benchmarks.
Conclusions:
- AntiBMPNN represents a significant advancement in antibody sequence design, offering superior accuracy and binding capabilities.
- The framework's ability to design high-affinity antibodies, validated experimentally, underscores its potential to revolutionize therapeutic antibody development.
