Protein structure prediction in structure-based ligand design and virtual screening

Advances in protein modeling algorithms and state-of-the-art sequence similarity comparison and fold recognition methods, in combination with growing protein structure information, are facilitating "genome-to-drug lead" approaches in which chemicals are virtually screened against computationally-predicted protein targets. Although the quality of predicted protein structures by homology modeling methods, and thus their applicability to drug discovery initiatives, predominantly depends on the sequence similarity between the protein of known structure and the protein target to be modeled, recent research underscores that this approach can be used to significant advantage in the identification and optimization of lead compounds, as well as for the identification and validation of drug targets. Rational structure-based drug design cycles begin with an iterative procedure that is dependent on the initial determination of the structure of the target protein, followed by the prediction of ligands for the target protein from molecular modeling computation. The structure determination of all proteins encoded by vast genome sequencing efforts appears to be an unrealistic goal with current technologies. Therefore, other approaches based on the development of technology useful for accurately predicting and modeling the structures of proteins have become exceedingly important in certain structure-based drug design efforts. This review provides an overview of the recent method advancements in protein structure prediction by homology modeling and includes an assessment of the application of homology modeling to pharmaceutically relevant questions. In addition, examples of successful applications of homology modeling approaches to genome-to-drug lead investigations are described.