Determination of Protein Tertiary Structure Class from Circular Dichroism Spectra

摘要: Abstract Fifty-three circular dichroism (CD) spectra consisting of the 46 native proteins, 3 denatured and one oligopeptide (the two proteins were taken at different temperatures) investigated in order to examine correlation between shape CD spectrum tertiary structure class protein. Five classes considered-all −α, all −β, α + β, α/β, proteins. Spectra from 190 236 nm with 2 interval described as points 24-dimensional hyperspace, where coordinates values ellipticities fixed wavelengths. This allows be treated patterns subsequently analyzed using pattern recognition algorithms. Cluster analysis, which does not need predefined information about protein structure, divides into several compact groups or clusters good class. To visualize these results, orthogonalization procedures imposed on original data set space. The new 3-dimensional coordinate system demonstrated well-separated all-β Regions corresponding −α especially β α/β well resolved. following approach was then applied obtain an objective mathematical algorithm for determination a protein′s its spectrum. hyperspace found by calculating decision functions, equations hyperplanes, separate spectral classes. representing region involves can interpreted this accuracy method checked removing training set, finding determinating excluded test gives 100% proteins; 85% 75% −β