Method and apparatus for predicting protein structure

A method of determining predicting putative contacting sites of a target protein is disclosed. The method uses a substitution matrix for calculating contact probabilities for pairs of columns in a sequence alignment containing the amino acid sequence of the target protein. The contact probabilities can also be used for predicting the three-dimensional structure of the protein. The putative contacts and/or three-dimensional structure typically correspond to the hydrophobic core of the target protein.