Use of advanced chemometrics for study of acidobasic and structural processes of bioligands

                                            Pavel Buček

The dissertation to be presented focuses on use of multivariate methods for data analyses (so
called advanced chemometrics). The data in form of both spectral curves and combination of discrete
points were obtained from various experiments monitored with optical spectroscopy methods. Use of
soft model and hard model based curve resolution methods is demonstrated on analysis of
experimental data of solution equilibria of two groups of oligonucleotides that are investigated in
relation to oncologic disorders. Acid-base properties of adenine analogues were studied by the same
means after we realized that there is scarce information about them despite their wide use in
therapeutic and research applications.

Use of artificial neural networks for pattern recognition of sensor arrays is discussed as another
approach to multivariate data analysis. Two arrays based on non specific cross-reactive sensors
were designed for two different groups of analytes as first step of an attempt to develop high
throughput sensor array for similar molecules. The feasibility of the concept is currently being
assessed. As the work is on the very beginning only problems encountered during the development of
data analysis procedure and steps taken towards their solution are described.