Year: 2016
Type: dissertation
Abstract: Modern, high-throughput techniques for the acquisition of metabolomic \ndata, combined with an increase in computational power, have provided \nnot only the need for, but also the means to develop and use, methods for \nthe interpretation of large and complex datasets. This thesis investigates \nthe methods by which pertinent information can be extracted from nontargeted \nmetabolomic data and reviews the current state of chemometric \nmethods. The analysis of real-world data and research questions relevant \nto the agri-food industry reveals several problems for which novel solutions \nare proposed. Three LC-MS datasets are studied: Medicago, Alopecurus \nand aged Beef, covering stress resistance, herbicide resistance and product \nmisbranding. The new methods include preprocessing (batch correction, \ndata-filtering), processing (clustering, classification) and visualisation and \ntheir use facilitated within a flexible data-to-results pipeline. The resulting \nsoftware suite with a user-friendly graphical interface is presented, providing \na pragmatic realisation of these methods in an easy to access workflow.
Author: Martin Rusilowicz
Language: English
Subfield: Molecular Biology
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W2763405510