Related Experiment Video
Updated: Jun 23, 2025
16:41
A Protocol for Computer-Based Protein Structure and Function Prediction
Published on: November 3, 2011
68.6K
EGG: Accuracy Estimation of Individual Multimeric Protein Models Using Deep Energy-Based Models and Graph Neural
Andrew Jordan Siciliano1, Chenguang Zhao2, Tong Liu1
1Department of Computer Science, University of Miami, 1365 Memorial Drive, Coral Gables, FL 33124, USA.
International Journal of Molecular Sciences
|June 19, 2024
Summary
We developed EGG (energy and graph-based architectures), a novel method to assess multimeric protein model accuracy. EGG ranked highly in CASP15 for predicting protein fold and interface quality.
Area of Science:
- Computational Biology
- Structural Biology
- Bioinformatics
Background:
- Accurate estimation of protein model quality is crucial for understanding protein function and biological processes.
- Assessing the accuracy of multimeric protein models, particularly their quaternary structure, is computationally challenging due to the vast conformational space.
- Existing methods struggle with the complexity of protein-protein interactions in multimeric complexes.
Purpose of the Study:
- To present EGG (energy and graph-based architectures), a new computational method for evaluating the accuracy of predicted multimeric protein models.
- To improve the assessment of both overall fold and interface accuracy in protein complexes.
- To provide a reliable tool for discerning the quality of protein quaternary structure predictions.
Main Methods:
- Development of EGG, utilizing message-passing and transformer layers.
- Implementation of graph-based architectures to infer structural properties.
- Evaluation of EGG's performance on CASP15 targets for predicting multimeric protein model accuracy.
Main Results:
- EGG achieved competitive results in CASP15, ranking fourth and third for overall fold and interface accuracy, respectively, for single highest-quality model prediction.
- EGG secured first place in CASP15 for identifying the top three highest-quality models concerning both overall fold and interface accuracy.
- Demonstrated superior performance compared to existing single model predictors in assessing multimeric protein models.
Conclusions:
- EGG provides a robust and accurate method for assessing the quality of predicted multimeric protein models.
- The developed approach significantly advances the computational assessment of protein quaternary structures.
- EGG's performance in CASP15 highlights its potential to enhance our understanding of protein complexes in various biological contexts.
