Gaussian-based alignment of protein structures: Deriving a consensus superposition when alternative solutions exist

The use of a Gaussian-based representation of protein structures for evalua ting protein-structure similarities and deriving three-dimensional superpos itions is presented. The approach, as implemented in the program GAPS, is a pplied to three pairs of proteins with different topological characteristic s (rich alpha-helix, mixed alpha-helix/beta-strand, and rich beta-strand), low sequence identities (10-30%), and recognized difficulties to define a u nique optimum alignment. Validation of the GAPS superpositions is done by c omparison with superpositions obtained by the TOP, GA_FIT, and ALIGN progra ms and those directly extracted from the FSSP database. Results suggest tha t a Gaussian-based methodology offers an objective means to, depending on t he Gaussian-based representation, derive a consensus three-dimensional supe rposition when alternative superposition solutions exist.