Differential Chemometric Analysis of Complex Mixtures

The following thesis used chemometric software to examine data generated from differential analysis of complex mixtures. Three different chemometric projects were investigated along with the comparison of two different chemometric software packages.

1. Liquid chromatography linked with mass spectrometry (LC-MS) was used to analyse gelatin from different species of animals. Using Agilent Technologies MPP chemometric software to analyse the data it was possible to find peptide fragments that were specific to each species of gelatin. Identification of these peptides was very difficult due to the destructive nature of gelatin manufacture. The method developed in this thesis allowed identification of 21 unknown gelatin samples with 100% accuracy.

2. Whiskey samples that had been matured in either American or Spanish oak barrels, and sampled over 22 years were analysed by LC-MS, the data was analysed with Agilent’s MPP chemometric software and the differences between the two wood types evaluated. Out of the compounds discovered to differentiate the whiskey from the different wood types two were identified as Gallic and Ellagic acid. These two compounds could be used to determine the wood type, a particular year of whiskey, or years of maturation for a whiskey matured in a particular wood type. The data generated for Whiskey maturity was also analysed using Waters Progenesis software and a comparison of the discovered compounds drawn. This highlighted an issue with incorrect assignment of neutral masses by MPP software. The data extraction method used by Progenesis was also found to be much less labour intensive than data extraction with MPP.

3. MPP was used to compare LC-MS data generated from the blood plasma of people suffering with multiple sclerosis with an age and gender matched control group. This was an un-targeted approach which led to the discovery of two compounds, Sphingosine and Dihydrosphingosine that were found to be lower in the blood of people suffering from multiple sclerosis. These compounds were searched in a larger sample set and found to follow the same trend of being lower in the disease group. It may be possible to use the concentration of these compounds in the blood as a marker of the disease.