Related Experiment Video
Updated: May 30, 2026
16:41
A Protocol for Computer-Based Protein Structure and Function Prediction
Published on: November 3, 2011
Prediction of conformational B-cell epitopes from 3D structures by random forests with a distance-based feature
Wen Zhang1, Yi Xiong, Meng Zhao
1School of Computer, Wuhan University, Wuhan 430072, China. zhangwen@whu.edu.cn
BMC Bioinformatics
|August 18, 2011
Summary
This study introduces a new bioinformatics method to accurately predict B-cell conformational epitopes from 3D structures, improving epitope-based drug design. The developed tool enhances the identification of key immune system binding sites.
Area of Science:
- Bioinformatics
- Immunology
- Structural Biology
Background:
- Antigen-antibody interactions are crucial for immune responses.
- B-cell epitopes are specific antigen sites recognized by antibodies.
- Accurate prediction of conformational epitopes is vital for epitope-based drug design but remains challenging.
Purpose of the Study:
- To develop a high-accuracy computational method for predicting B-cell conformational epitopes.
- To address limitations in existing conformational epitope prediction methods.
Main Methods:
- Introduced a 'thick surface patch' concept to incorporate interior residue information.
- Developed an 'adjacent residue distance' feature to capture unequal residue contributions.
- Utilized a bootstrapping and voting procedure to handle imbalanced datasets.
- Employed the random forest algorithm for classification.
Main Results:
- The proposed method demonstrates high accuracy in predicting conformational B-cell epitopes from 3D structures.
- Achieved mean AUC values of 0.633 (bound) and 0.654 (unbound) datasets.
- Outperformed or matched state-of-the-art prediction models in independent tests.
Conclusions:
- The developed method is effective for predicting conformational epitopes.
- A publicly available tool has been created to predict conformational epitopes from 3D structures.
