Application of cluster analysis to high-throughput multiple data types

Cunningham, Gordon John (2011) Application of cluster analysis to high-throughput multiple data types. PhD thesis, University of Glasgow.

Abstract

PolySNAP is a program used for analysis of high-throughput powder diffraction data. The
program matches diffraction patterns using Pearson and Spearman correlation coefficients
to measure the similarity of the profiles of each pattern with every other pattern, which
creates a correlation matrix. This correlation matrix is then used to partition the patterns
into groups using a variety of cluster analysis methods. The original version could not
handle any data types other than powder X-ray Diffraction. The aim of this project was to
expand the methods used in PolySNAP to allow it to analyse other data types, in particular
Raman spectroscopy, differential scanning calorimetry and infrared spectroscopy data.
This involves the preparation of suitable compounds which can be analysed using these
techniques. The main compounds studied are sulfathiazole, carbamazepine and piroxicam.
Some additional studies have been carried out on other datasets, including a test on an
unseen dataset to test the efficacy of the methods.
The optimal method for clustering any unknown dataset has also been determined.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Subjects: Q Science > QD Chemistry
• Colleges/Schools: College of Science and Engineering > School of Chemistry
• Supervisor's Name: Gilmore, Prof. Chris
• Date of Award: 2011
• Depositing User: Mr Gordon Cunningham
• Unique ID: glathesis:2011-2715
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 22 Jun 2011
• Last Modified: 10 Dec 2012 13:59
• URI: https://theses.gla.ac.uk/id/eprint/2715