Structural bioinformatics: from protein structure to function

A major problem faced by structural biology today is the issue of function prediction. With the success of the various Structural genomics initiatives and advances in crystallography, proteomics, and other experimental techniques, there has been an explosion of new protein structures being deposited in the databases. In many cases, however, these proteins have little or no functional annotation. Sequence-based approaches still remain the most effective way to assign function based on homology, but in cases of extreme divergence and analogous proteins these methods can fall. In order to identify these types of relationships, a number of structure-based approaches have been developed, such as the MSDmotif service. No single method is successful in all cases and a more prudent approach involves the utilization of data from a wide range of resources. One such approach is the ProFunc server, developed to help researchers narrow down the number of functional possibilities for experimental validation.