Genetically evolved receptor models: a computational approach to construction of receptor models

Area of Science:

• Computational chemistry
• Drug discovery
• Structural biology

Background:

• Designing ligands for receptor sites often requires known 3D receptor structures, which are frequently unavailable.
• Existing methods for ligand design are limited when receptor structural data is absent.

Purpose of the Study:

• To develop a computational method for modeling receptor sites without prior 3D structural information.
• To utilize genetic algorithms and structure-activity relationships (SAR) for de novo receptor modeling.

Main Methods:

• The GERM program employs a genetic algorithm to generate atomic-level models of receptor sites.
• Models are based on a limited set of known structure-activity relationships.
• Intermolecular energies are calculated to correlate with bioactivities.

Main Results:

• Evolved receptor models demonstrate a strong correlation between calculated intermolecular energies and observed bioactivities.
• The models provide reliable predictions of bioactivity for compounds not used in the initial model generation.
• Generated models serve as effective starting points for subsequent ligand design efforts.

Conclusions:

• The GERM program offers a viable computational strategy for modeling receptor sites when 3D structures are unknown.
• This approach facilitates both computational and human-driven ligand design processes.
• The method enhances drug discovery pipelines by enabling modeling from SAR data alone.