Data analysis of chromatographic fingerprints.

Chromatographic fingerprinting is rapidly emerging as the tool of interest for the analysis of samples with a complex composition. By spreading the chemical information over time, the individual compounds are revealed, resulting in a characteristic profile. Based on this information, quality control criteria are gradually redefined and higher standards are set. However, the abundance of data generated also introduces new challenges to the field. In this context, the thesis deals with the extraction of the relevant information from the fingerprints. Based on the interaction between the data and chemometric tools, the differentiation of related herbal species is addressed. The constructed discrimination models also allow identifying potential marker compounds which can be used to update the currently defined quality control criteria. Additionally, the pharmacological activity of natural products is linked to their fingerprints, indicating the compounds potentially responsible for the measured activity. Besides the applications, the thesis also addresses some fundamental challenges in the data analysis of fingerprints. By introducing the available spectral data, bilinear resolution methods are further developed and implemented to cope with the complexity of the data. Their applicability is demonstrated resolving constituents with a high degree of overlap, while providing a solution for peak shifting and quantification issues.