Critical evaluation of the research docking program for the CASP2 challenge

Area of Science:

• Computational chemistry
• Structural biology
• Drug discovery

Background:

• Accurate prediction of small-molecule ligand binding poses is crucial for understanding protein-ligand interactions and advancing drug discovery.
• Computational docking programs play a vital role in predicting these binding poses, aiding in the identification of potential drug candidates.

Purpose of the Study:

• To evaluate the performance of the Research docking program in predicting ligand binding positions for protein complexes in the CASP2 challenge.
• To assess the accuracy of the program's energy function and Out-rank scoring system in ranking correct binding poses.

Main Methods:

• Utilized the Research docking program, employing a Monte Carlo procedure with pairwise energies for conformational searching of ligand torsional space.
• Predicted binding positions for six small-molecule ligands complexed with target proteins.
• Evaluated predictions using root-mean-square (RMS) deviation from experimental structures and ranking by interaction energy or Out-rank score.

Main Results:

• Successfully predicted 2 out of 5 noncovalent complexes within 2 Å RMS of experimental structures, based on interaction energy or Out-rank score.
• Achieved a correct docking within 2 Å RMS for 4 out of 5 noncovalent structures within the top 20 ranked poses by energy.
• Identified limitations in discriminating among the lowest energy dockings due to the energy function and Out-rank scoring.

Conclusions:

• The Research docking program demonstrates encouraging success in exploring the multi-dimensional docking space (position, orientation, conformation).
• Further refinement of the energy function and scoring mechanism is needed to improve discrimination among top-ranked poses.
• The program shows potential for accurate prediction of ligand-protein interactions in computational drug design.