Automated processing and analysis of gas chromatography/mass spectrometry screening data
AuthorsHitchcock, Jonathan James
sports drug surveillance
B290 Pharmacology, Toxicology and Pharmacy not elsewhere classified
F151 Pharmaceutical Chemistry
MetadataShow full item record
AbstractThe work presented is a substantial addition to the established methods of analysing the data generated by gas chromatography and low-resolution mass spectrometry. It has applications where these techniques are used on a large scale for screening complex mixtures, including urine samples for sports drug surveillance. The analysis of such data is usually automated to detect peaks in the chromatograms and to search a library of mass spectra of banned or unwanted substances. The mass spectra are usually not exactly the same as those in the library, so to avoid false negatives the search must report many doubtful matches. Nearly all the samples in this type of screening are actually negative, so the process of checking the results is tedious and time-consuming. A novel method, called scaled subtraction, takes each scan from the test sample and subtracts a mass spectrum taken from a second similar sample. The aim is that the signal from any substance common to the two samples will be eliminated. Provided that the second sample does not contain the specified substances, any which are present in the first sample can be more easily detected in the subtracted data. The spectrum being subtracted is automatically scaled to allow for compounds that are common to both samples but with different concentrations. Scaled subtraction is implemented as part of a systematic approach to preprocessing the data. This includes a new spectrum-based alignment method that is able to precisely adjust the retention times so that corresponding scans of the second sample can be chosen for the subtraction. This approach includes the selection of samples based on their chromatograms. For this, new measures of similarity or dissimilarity are defined. The thesis presents the theoretical foundation for such measures based on mass spectral similarity. A new type of difference plot can highlight significant differences. The approach has been tested, with the encouraging result that there are less than half as many false matches compared with when the library search is applied to the original data. True matches of compounds of interest are still reported by the library search of the subtracted data.
PublisherUniversity of Bedfordshire
TypeThesis or dissertation
DescriptionA thesis submitted to the University of Bedfordshire, in partial fulfilment of the requirements for the degree of Ph.D.
The following license files are associated with this item: