Development of advanced chemometric methods for the analysis of deep-UV resonance Raman spectra of proteins

Abstract

[ACCESS RESTRICTED TO THE UNIVERSITY OF MISSOURI AT AUTHOR'S REQUEST.] Ultra violet resonance Raman (UVRR) is a powerful spectroscopic technique for determining the secondary structural content of proteins in solution. However, the analyses of UVRR spectra can be problematic due to the difficulty of determining the pure secondary structure Raman spectra. The use of multi-excitation datasets can help to alleviate the difficulty in determining the pure secondary structure Raman spectra, but due to increasing spectral resolution as excitation wavelength increases, multi-excitation datasets are notoriously difficult to align spectral features. In addition, the subtraction of the water band can be difficult when relying on an internal intensity standard. To address these difficulties we demonstrate the use of a series of chemometric methods. To determine the pure secondary structure Raman spectra, we demonstrate the use of multivariate curve resolution using the alternating least squares algorithm (MCRALS) with a multi-excitation data set. To alleviate mis-alignment in multi-excitation data, we demonstrate the use of correlation optimized warping (COW). We also propose a new water band subtraction method which will reliably determine the water band concentration and remove it, without an over subtraction. Finally, we demonstrate the use of parallel factor analysis (PARAFAC) for the characterization of the behavior of the flavonoid quercetin in solution with the protein bovine serum albumin.