The goal of protein modeling is to predict a structure from its sequence with an accuracy that is comparable to the best results achieved experimentally. This would allow users to safely use in silico generated protein models in scientific fields where today only experimental structures provide a solid basis: structure-based drug design, analysis of protein function, interactions, antigenic behavior, or rational design of proteins with increased stability or novel functions. Protein modeling is the only way to obtain structural information when experimental techniques fail. Many proteins are simply too large for NMR analysis and cannot be crystallized for X-ray diffraction.
Among the three major approaches to 3D structure prediction described in this and the following two chapters, homology modeling is the “easiest” approach based on two major observations:
- The structure of a protein is uniquely determined by its amino acid sequence (Epstain, Goldberger, and Anfinsen, 1963), and therefore the sequence should, in theory, contain suffice information to obtain the structure.
- During evolution, structural changes are observed to be modified at a much slower rate than sequences. Similar sequences have been found to adopt practically identical structures while distantly related sequences can still fold into similar structures. This relationship was first identified by Chothia and Lesk (1986) and later ...