COMPARISON OF THE LINEAR LEAST MEAN SQUARES PROCEDURE AND THE VECTOR PROJECTION METHOD OF SPECTRAL ANALYSIS

A mathematical analysis of two methods commonly used for the determination of fractions of secondary structure from circular dichroism spectral data-the vector projection method of Baker & Isenberg [Biochem. 15, 629 (1976)] and the classical linear least mean squares procedure-reveals that the two procedures handle errors of all types in an equivalent fashion and yield identical values of the fractional amounts of secondary structure present, as long as the same method of integration of the data is employed. Analyses of simulated data show that the method of integration [rectangular, trapezoidal or parabolic (Simpson's Rule)] has an insignificant effect on the calculated values of the fractions of secondary structure present. Our study also shows that the insensitivity of the fraction of α helix calculated using a given random coil reference spectrum is a natural consequence of certain intrinsic properties of the reference spectra of the α helix, β structures and random coil forms, rather than the method of calculation employed. © 1979.