Related Experiment Video
Updated: Jun 4, 2025
12:28
Bacterial Inner-membrane Display for Screening a Library of Antibody Fragments
Published on: October 15, 2016
11.5K
De Novo Antibody Design with SE(3) Diffusion
Daniel Cutting1, Frédéric A Dreyer1, David Errington1
1Exscientia, Oxford Science Park, Oxford, UK.
Summary
IgDiff, a novel antibody variable domain diffusion model, generates highly designable antibodies with unique binding regions. Experimental validation confirms high expression yields for these designed antibodies.
Area of Science:
- Computational biology
- Protein engineering
- Immunoinformatics
Background:
- Antibody design is crucial for therapeutic development.
- Existing generative models face limitations in antibody structure generation and designability.
- Developing advanced computational tools is essential for novel antibody discovery.
Purpose of the Study:
- Introduce IgDiff, a diffusion model for antibody variable domain generation.
- Evaluate the designability and novelty of antibodies generated by IgDiff.
- Compare IgDiff's performance against state-of-the-art models in antibody design tasks.
Main Methods:
- Adapted a general protein backbone diffusion framework to handle multiple antibody chains.
- Assessed generated antibody structures for designability and novelty.
- Performed experimental verification of designed antibody expression and yield.
- Benchmarked IgDiff against a leading generative backbone diffusion model.
Main Results:
- IgDiff successfully generates highly designable antibodies, including novel binding regions.
- Sampled antibody structures exhibit good agreement with reference antibody backbone dihedral angle distributions.
- All experimentally verified designed antibodies expressed with high yield.
- IgDiff demonstrated improved properties and designability over a state-of-the-art model in CDR design and chain pairing tasks.
Conclusions:
- IgDiff represents a significant advancement in computational antibody design.
- The model's ability to generate novel and designable antibody structures holds promise for therapeutic antibody development.
- IgDiff offers a powerful tool for various antibody engineering applications, including complementarity determining region design and antibody chain pairing.
