ALIGNMENT OF CHROMATOGRAPHIC PROFILES FOR PRINCIPAL COMPONENT ANALYSIS - A PREREQUISITE FOR FINGERPRINTING METHODS

In fingerprinting methods, small differences between chromatograms with rather complex appearance have to be detected. Pattern recognition methods based on principal component analysis (PCA) could be a useful tool, but for chromatographic profiles as input data a severe problem is the great impact of chromatographic variations compared with true variations in sample composition. The problem has been analysed in terms of parameter variations for exponentially modified Gaussian peaks, and a procedure has been developed to align a sample chromatogram towards a target chromatogram in order to compensate for (i) small shifts in retention time (not due to different sample components), (ii) common variations in peak area (not due to sample composition) and (iii) variations in level and slope of the baseline. The effects of the alignment procedure on the PCA is demonstrated for a set of chromatographic profiles intended for peptide mapping.