INFERRING PROTEIN FUNCTION FROM STRUCTURE
The Importance of Predicting Protein Function from Structure
The various genome-sequencing projects around the globe have provided us with massive amounts of information detailing all the genes in a number of organisms required for survival. The amount of sequence data is set to rapidly expand as a result of the very large-scale metagenomics projects such as the Global Ocean Survey (Yooseph et al., 2000). By comparison, the protein structure data falls far behind. Structural genomics aims to close the gap by experimentally determining a large number of protein structures as rapidly and accurately as possible using high-throughput methods. There are several consortia working on this across the globe and each has its individual goals, but one of the key aims is to increase the coverage of protein fold space and hence the proportion of protein sequences amenable to homology modeling methods. As a result, an increased number of protein structures have been released with little or no functional annotation. This is a reversal of the usual experimental investigation of proteins, which involves taking a protein of interest, carrying out biochemical experiments to determine functional information about it, and then using the structure to rationalize this functional information (Thornton et al., 2000). For example, the tyrosine kinases were known to be signaling molecules long ...