Related Experiment Video
Updated: Sep 9, 2025
06:50
Author Spotlight: A Computational Approach to Decipher Amino Acid Preferences in Multispecific Protein-Protein Interactions
Published on: January 26, 2024
2.0K
RESP2: An Uncertainty Aware Multi-Target Multi-Property Optimization AI Pipeline for Antibody Discovery
Jonathan Parkinson1,2, Ryan Hard1, Young Su Ko1
1Department of Chemistry and Biochemistry, University of California San Diego, La Jolla, CA, 92093-0359, USA.
Advanced Science (Weinheim, Baden-Wurttemberg, Germany)
|September 5, 2025
Summary
We developed RESP2, a new computational pipeline for discovering therapeutic antibodies. It efficiently finds antibodies that bind strongly to multiple target variants, outperforming current AI methods.
Area of Science:
- Computational biology
- Immunology
- Protein engineering
Background:
- Therapeutic antibody discovery for infectious diseases faces challenges due to pathogen mutation.
- Antibodies require high affinity for multiple antigen variants and optimal developability properties.
Purpose of the Study:
- To introduce RESP2, an enhanced computational pipeline for discovering therapeutic antibodies.
- To enable simultaneous optimization of antibody binding affinity and developability against evolving targets.
Main Methods:
- In silico evaluation of the RESP2 pipeline using the Absolut! antibody-antigen docking database.
- Case study using the COVID-19 spike protein receptor binding domain (RBD) and its variants.
Main Results:
- RESP2 identified antibody sequences with significantly higher binding affinity to target antigen groups compared to training data (≥ 85% success rate).
- RESP2 outperformed popular generative AI techniques, achieving success rates ≤ 1.5%.
- A novel human antibody was discovered with mid-to-high affinity binding to at least eight COVID-19 RBD variants.
Conclusions:
- The RESP2 pipeline offers a powerful advantage for discovering antibodies against rapidly evolving infectious disease targets.
- RESP2 facilitates the development of broadly protective therapeutic antibodies.
- A publicly available Python package for RESP2 is provided for broader research use.
