Related Experiment Videos
A knowledge-based energy function for protein-ligand, protein-protein, and protein-DNA complexes
Chi Zhang1, Song Liu, Qianqian Zhu
1Howard Hughes Medical Institute Center for Single Molecule Biophysics, Department of Physiology & Biophysics, State University of New York at Buffalo, 124 Sherman Hall, Buffalo, New York 14214, USA.
Journal of Medicinal Chemistry
|April 2, 2005
Summary
A new knowledge-based statistical energy function, DFIRE, accurately predicts binding affinities for protein-ligand, protein-protein, and protein-DNA complexes. It shows strong correlations with experimental data, outperforming other scoring functions in benchmarks.
Area of Science:
- Computational Biology
- Structural Bioinformatics
- Biophysics
Background:
- Accurate prediction of molecular binding affinities is crucial for drug discovery and understanding biological interactions.
- Existing scoring functions often struggle to generalize across different types of molecular complexes.
Purpose of the Study:
- To develop and validate a knowledge-based statistical energy function, DFIRE, for predicting binding affinities in protein-ligand, protein-protein, and protein-DNA complexes.
- To compare the performance of DFIRE against other established scoring functions.
Main Methods:
- Developed a knowledge-based statistical energy function using 19 atom types and a distance-scale finite ideal-gas reference (DFIRE) state.
- Trained and tested the DFIRE function on experimentally measured binding affinities for various molecular complexes.
- Benchmarked DFIRE against 12 other scoring functions using protein-ligand docking decoys.
Main Results:
- DFIRE achieved correlation coefficients of ~0.63 for protein-ligand binding affinities.
- Demonstrated high accuracy for protein-protein (0.73) and protein-DNA (0.83) binding affinities, even without specific training on these complex types.
- Outperformed other scoring functions in correlating theoretical and experimental binding affinities and in ranking docking decoys.
Conclusions:
- The DFIRE energy function provides a robust and accurate method for predicting binding affinities across diverse molecular complex types.
- DFIRE shows significant potential for applications in structural bioinformatics and drug discovery.
- The DFIRE energy function parameters and program are available for academic use.