Hydrophobic neighboring homology (HNH) dotplot: an approach for assessing structurally similar motifs in proteins.

Structural biology needs sensitive tools to detect homology between proteins of low sequence identity, but with closely related 3-D structures. Using a conventional dotplot method, we therefore introduced 2 concepts to improve the search for similarities between secondary structures of analyzed proteins: 'hydrophobic neighboring homology' (HNH) and 'amino acid degeneracy classes'. The amino acids are grouped into 3 subsets: hydrophobic, hydrophilic and mimetic. A 'Neighboring Similarity Index' (NSI) is calculated for every residue pair and quantifies its neighbor homology. By thresholding the homology matrix and filtering the dotplot, the homologous patterns are extracted. We have evaluated the efficiency and limits of the method using 21 protein pairs extracted from the Protein Data Bank (PDB), or selected from the recent literature. Globally, we again find the homologous structures (alpha-helices and beta-strands) of these pair proteins. The introduction of neighbor residue hydrophobicity in the conventional dotplot improves the alignment of proteins with low sequence identity (< 25%). HNH, written in standard ANSI C with the graphic library X11, under UNIX, is available on request.