Objectively obtaining information from gas chromatographic separations of complex samples using novel data processing and chemometric techniques

Analysts often rely upon one dimensional, multidimensional, or hyphenated chromatography to obtain information about their sample of interest, but chromatography often produces large volumes of data with uncontrollable variations arising from sources other than chemical variations. Analysts must be able to objectively and efficiently process and reduce these large volumes of data to obtain the desired information. The research presented herein addresses this need through the development of objective data processing and chemometric techniques to improve retention time precision, reduce undesirable sources of variation, and classify or quantify complex samples. One novel tool is a fast, robust piecewise retention time alignment algorithm with objective parameter optimization that reduces retention time variations in chromatographic data sets while preserving chemical variations to benefit chemometric analysis. The piecewise alignment algorithm was further evaluated using simulated and real chromatographic data sets with declining retention time precision. This evaluation revealed that the piecewise alignment algorithm is capable of correcting severe retention time shifts. Once the piecewise alignment algorithm was found to be suitably robust and quick, it was adapted for comprehensive two-dimensional gas chromatography (GC x GC). The two-dimensional piecewise alignment algorithm improved the bilinearity of GC x GC separations. After researching novel standardization and classification techniques for the first and second order instrumental applications, third order instrumental data analysis techniques were developed. Comprehensive two-dimensional gas chromatography coupled with time-of-flight mass spectrometry (GC x GC-TOFMS) is capable of separating thousands of components in a complex sample, providing a wealth of data in the form of a threeway array. GC x GC-TOFMS separations of metabolite extracts were a novel application of unsupervised multiway PCA to determine the metabolic differences among the samples. Another third order data analysis tool, the Fisher ratio algorithm, was developed to identify statistically significant sample components that differentiated known classes of complex samples in a supervised manner that was robust against within-class variations. This algorithm was applied to the entire fourway array of multiple GC x GC-TOFMS separations of complex samples. Finally, a data processing and reduction technique for high-speed gas chromatography-mass spectrometry separations of complex samples was developed for improved pattern recognition.