Use of /spl beta/-strand interaction pseudo-potentials in protein structure prediction and modelling

A residue-residue interaction pseudo-potential has been developed specific for protein /spl beta/-sheets. The potential is derived by scoring the occurrence of all i-(j-2...J+2) residue-residue pairs between any two interacting /spl beta/-strands, subdividing according to 4 classes of hydrogen bond pattern. The potential can be useful in distinguishing between correct and incorrect alignments between /spl beta/-strands in a predicted protein structure. It is not unduly influenced by the quality or completeness of a model, because it is not based on atom-atom distances. The potential can also be used directly to predict a /spl beta/-strand contact map for any protein sequence. Maps generated when considering a single protein sequence have a poor signal to noise ratio, however when a prediction is smoothed by considering a family of homologous sequences, native and non-native /spl beta/-strand pairings can be partially distinguished.<>