Resonant Recognition Model Defines the Secondary Structure of Bioactive Proteins

摘要

The Resonant Recognition Model (RRM) of protein bioactivity is applied to the protein secondary structure prediction. The method is based on the physical and mathematical model of the electron-ion interaction pseudopotential (EIIP) and uses signal analysis to interpret linear information contained in a macromolecular sequence. The method of analysis is based on a two-step procedure. Protein sequence is first transformed into a numerical series by means of the individual EIIP amino acid values. The second step of the model involves the Fourier spectral analysis of the obtained numerical series. Cosic et al. have shown that single frequency peaks of the spectrum define characteristic positions of the amino acids, i.e., hot spots, correlated to the biological function of the protein. We have analysed the secondary protein structure by comparing the patterns of 20 most prominent frequency peaks of the single-series Fourier RRM periodogram. The patterns within 140 nonhomologous α- and β(3-protein folds obtained from the Jpred and SCOP databases were analysed by means of the classification tree in order to obtain the algorithm for the α- and β-fold classification. This quick and simple procedure of the secondary fold prediction showed high accuracy of 98.55%. The stability of the tree algorithm solution was confirmed by jack-knife testing of the tree algorithm (mean error 2.6). This method of the secondary structure predic- tion is presented in more detail on a subset of 30 different cyto-kines, hormones, enzymes and viral proteins. Our results indicate that resonant spectral analysis of the protein primary amino acid sequence may be used to extract information about its secondary structure.