Large-Scale Comparative Protein Modelling

Proteins are the working molecules in most biological processes and a large portion of experimental biology is aimed at understanding their function. To understand function it is necessary to know the shape of the protein and where or how it interacts with other proteins or substrates. This is approached by first solving the three-dimensional (3-D) structure and then changing the protein sequence to see how the structure and/or function is altered. Site-directed mutagenesis experiments are generally designed to provide information about the involvement of a protein’s specific residues in enzyme-based reactions, molecular recognition events, protein stability and during drug design projects. The experimental elucidation of 3-D structures by X-ray crystallography or NMR is, however, often hampered by difficulties in obtaining sufficient pure protein, diffracting crystals, the size of the protein or its subunits for NMR studies and many other technical aspects. This is illustrated by the fact that the SWISS-PROT/TrEMBL database (Bairoch and Apweiler 1997) contains around 175,000 sequences, while the Brookhaven Protein Data Bank (PDB) (Abola et al. 1996) contains about 6,000 3-D structures from 1,700 different proteins. The number of solved 3-D structures is increasing very slowly compared to the rate of sequencing of novel cDNAs, and no structural information is available for the vast majority of known protein sequences. This gap will of course, further increase as genome sequencing projects continue to yield large numbers of novel sequences.