Modelling by homology of the HSV1-TK sequence embedded structural alignment.

Abstract

Modelling by homology is an approach to the rational design of new drugs based on the construction of ligand protein interaction complexes. Because in most cases the 3D-structure of the target protein is not known from biophysical data, this approach yields a theoretical procedure which establishes at least parts of the protein by comparison with isofunctional proteins, assuming that much of the structural information is embedded in the amino acid sequence. This approach should be of considerable importance for proteins with divergent primary structures but with a high degree of isofunctionality, the latter demanding a similar active site folding pattern. This study is a pattern recognition approach based on additive secondary structure prediction and surface probabilities from residue variabilities. The comparison of the additive properties yields a sequence alignment of the viral thymidine kinases with the adenylate kinases having a closely related functionality. X-ray structures of adenylate kinases can then be used as templates to derive a 3D-structure prediction of the thymidine kinase active site.