Methods of clustering proteins

The invention relates to a method of clustering a set of proteins based on the sequence similarities of functional domain(s) such as pocket(s), functional site(s), allosteric site(s), and active site(s). The functional domain(s) of a protein sequence are identified based on the three-dimensional structure of the protein. Proteins are clustered based on the sequence similarity of the amino acid residues of the functional domain(s) and represented as a dendrogram. Proteins in a particular cluster show similar interaction patterns with specific drugs. Methods for identifying modulators for drug discovery based on the similarities of the functional domain(s) are provided.