Related Experiment Videos
SMall Molecule Growth 2001 (SMoG2001): an improved knowledge-based scoring function for protein-ligand interactions.
Alexey V Ishchenko1, Eugene I Shakhnovich
1Department of Chemistry and Chemical Biology, Harvard University, 12 Oxford Street, Cambridge, MA 02138, USA.
Journal of Medicinal Chemistry
|June 14, 2002
Summary
We developed SMoG2001, an improved scoring function for computational drug design. It accurately predicts ligand-protein binding affinities, outperforming previous versions and other leading functions.
Area of Science:
- Computational chemistry
- Structural biology
- Drug discovery
Background:
- Accurate scoring functions are crucial for ranking virtual ligands in computational drug design.
- Existing functions face limitations in speed and precision for large-scale virtual screening.
Purpose of the Study:
- To introduce SMoG2001, an enhanced knowledge-based scoring function for predicting protein-ligand binding affinities.
- To evaluate SMoG2001's performance against established scoring functions.
Main Methods:
- Utilized a knowledge-based approach, deriving free energy parameters from atom-atom contact frequencies in 725 protein-ligand complexes.
- Employed statistical mechanics principles to establish the reference state (no interactions) for improved accuracy.
Main Results:
- SMoG2001 demonstrated good accuracy in reproducing experimental binding constants for 119 test complexes.
- Outperformed PMF and SCORE1(LUDI) and showed comparable results to DrugScore on similar test sets.
- Identified limitations in predicting affinities for large, flexible ligands and those involving quantum mechanical interactions with metal ions.
Conclusions:
- SMoG2001 offers improved accuracy in predicting binding affinities compared to previous SMoG versions, attributed to a refined reference state description.
- The function shows promise for computational lead design but requires further development for specific ligand types.