Screening drug-like compounds by docking to homology models: a systematic study

Area of Science:

• Computational chemistry and structural biology
• Drug discovery and development

Background:

• Homology modeling is a crucial tool for drug discovery when experimental protein structures are unavailable.
• Virtual screening using docking and scoring relies on accurate protein models.

Purpose of the Study:

• To evaluate the effectiveness of homology models in virtual screening for drug targets.
• To investigate the impact of template choice, screening accuracy, and protein flexibility on docking results.

Main Methods:

• Generated multiple homology models for five different drug targets.
• Performed virtual screening calculations using docking and scoring on these models.
• Assessed screening enrichment and compared results with docking to crystal structures and templates.

Main Results:

• Docking to homology models achieved comparable ligand enrichment to docking with actual crystal structures.
• Template similarity showed a weak correlation with screening performance; docking to templates was often as effective as to homology models.
• Allowing for protein side-chain flexibility offered a modest improvement in screening outcomes.

Conclusions:

• Homology models are a viable alternative to experimental structures for virtual screening in drug discovery.
• The choice of template and protein flexibility have less impact than anticipated, suggesting simplified modeling approaches may suffice.
• Further research is needed to understand the underlying reasons for these findings and refine virtual screening methodologies.