Related Experiment Video
Updated: Jun 7, 2026
05:19
Rapid Glyco-Qualitative Assessment of Recombinant Proteins Using a Fully Automated System
Published on: June 28, 2024
LeGenD: High-throughput N-glycan profiling using explainable AI and lectin profiling
Haining Li1, Angelo G Peralta2, Sanne Schoffelen3
1Department of Bioengineering, University of California, San Diego, La Jolla, California, USA.
The Journal of Biological Chemistry
|June 5, 2026
Summary
A new AI platform called LeGenD uses lectin binding to predict N-glycan structures on proteins. This method offers a faster, cost-effective alternative for analyzing protein glycosylation patterns.
Area of Science:
- Biochemistry
- Glycobiology
- Computational Biology
Background:
- Protein glycosylation is crucial for biological functions, impacting areas from basic research to biopharmaceutical development.
- Conventional glycan analysis methods face challenges in throughput and cost.
- Lectins provide glycan epitope information but lack full structural details.
Purpose of the Study:
- To develop an AI-driven platform, LeGenD, for predicting dominant N-glycan structures and their abundance on proteins.
- To overcome limitations of existing glycan analysis techniques.
Main Methods:
- LeGenD integrates lectin-binding patterns with artificial intelligence (AI) to predict N-glycan structures.
- The model was trained on glycoprofiles from recombinant proteins produced in glycoengineered CHO cell lines.
- SHapley Additive exPlanations (SHAP) were used to identify key lectins for prediction.
Main Results:
- LeGenD effectively predicts dominant glycosylation patterns on purified proteins.
- The approach demonstrated high accuracy on independent test data.
- Key lectins influencing glycoprofile predictions were identified.
Conclusions:
- LeGenD offers a novel, AI-based platform for analyzing protein glycosylation.
- This approach provides an alternative to conventional methods, potentially complementing existing glycan analysis toolkits.
