Prediction of biologically active regions in protein sequences via best basis selection

A common biological function of a set of protein sequences can be identified using the resonant recognition model (RRM), where frequency analysis of the protein sequences and peak detection are used. In this paper, local amino acid regions associated with a common biological function are predicted using space-frequency analysis of protein sequences. After a relative shift of each protein sequence, the highest-amplitude space-frequency tile of the best basis is used to predict biologically active regions. Joint processing of sequences using best basis selection provides fine spatial resolution in active region prediction, which can be useful for peptide synthesis and drug design.