Michael Levitt, Stanford University, Stanford, USA

Methods for Structural Alignment and their Application to Linking Protein Folds
Structural Alignment complements Sequence Alignment in that it compares protein three-dimensional structures rather than their one-dimensional sequences. It is a technique that is essential for studying the connections between proteins that result from evolution as sequence alignment is much less powerful than structural alignment. In other words structural alignment can detect significant similarities between proteins that are much more dissimilar than can be detected by sequence alignment. This greater sensitivity comes at a price: the powerful, O(n2) method of dynamic programming can be applied to sequence alignment, whereas structural alignment is at least O(n8). As a result there is just one basic method to compare sequences (albeit with many modifications which make the method faster), there are at least 50 methods to compare structures. We developed one of the first such methods, Structal, and have recently compared its performance to some of the other powerful and popular methods. This tutorial will describe the basic principles of structural alignment, explain its complexity, compare the performance of different methods and show how structural alignment aids automatic classification of protein folds.

Laurents, D.V., S. Subbiah and M. Levitt. (1994). Different Protein Sequences Can Give Rise to Highly Similar Folds Through Different Stabilizing Interactions. Protein Science 3 :1938-1944.

Kolodny R., Koehl P. and M. Levitt. (2005). Comprehensive evaluation of protein structure alignment methods: scoring by geometric measures. J. Mol. Biol . 346 : 1173-1188.