The analysis of Raman optical activity spectra of proteins
Measurement of the Raman optical activity (ROA) spectra of biomolecules has become an experimental possibility due to significant advances in the available technology, and its successful implementation into the ROA instruments at the University of Glasgow. The ease with which the ROA spectra of biological molecules can be successfully measured lends itself perfectly to the ever-growing demand for biomolecular structural information, especially in the context of proteomics and the Human Genome Project. ROA spectroscopy is able to probe the chiral peptide backbone of proteins, and as such the ROA spectrum of a protein contains a wealth of structural information from within the whole molecule, across the whole vibrational spectrum. As well as containing detailed information from specific structural elements such as sections of secondary structure and motifs, the ability of ROA to see the molecule as a whole also enables the global fold of the protein to be deduced from the ROA spectrum. The development of the analysis of ROA spectra has largely been based upon the correlation of ROA spectra of proteins of known structure with structural information from alternative sources, chiefly X-ray crystallography and multidimensional nuclear magnetic resonance (NMR). As the database of ROA spectra of polypeptides and proteins has grown, it has been possible to tighten up the assignment of ROA spectral bands and band patterns to aspects of known structural content. With a basis for the correlation between the ROA spectrum and the known crystal structure (or NMR structure) being well established, it is possible to interpret the ROA spectra of proteins that do not have (for whatever reason) well defined structures. This means that ROA spectroscopy can provide invaluable structural information for proteins that are precluded from analysis by other techniques, and also cast new light on the structures of proteins that have not been well defined. In order to fully interpret an ROA spectrum of a protein, it is necessary to be familiar with protein structure and the ROA experiment as a whole. Analysing an ROA spectrum is a detailed and highly subjective process. Depending on the experience of the analyst, the information contained within the spectra can be extracted readily or not so readily. For this reason, it would be desirable to develop a technique that is capable of interpreting not only individual spectra, but also whole data sets in a manner that is independent of the analyst, and therefore independent of any preconceptions (or inexperience) the analyst may have. This project presents an up-to-date collection of newly obtained ROA spectra of a large number of proteins across a range of structural class types. In addition, the statistical technique of principal component analysis (PflA) has been used as a tool for the analysis of this new data. It is hoped that the result of this work will provide a basis for the future analysis of protein ROA spectra that is both mathematically rigorous and convenient.