Evaluation of chemometric software for analysis of complex mixtures for biologically derived samples analysed using liquid-chromatography mass spectrometry

Scott, Megan Rose

Abstract

There is a risk of losing important information when choosing the best metabolomic workflow for untargeted chemometric analysis. Choosing the best software for conducting this analysis is crucial as wrong parameters with the wrong software could lead to false negatives, as well as over-saturating the data analysis with false positives. Over the course of this study, the intention was to show how a robust untargeted liquid chromatography-mass spectrometry method, followed by deconvolution then performing statistical analysis can determine consistent, concise and accurate markers that explain differences between datasets. Different software packages were used throughout to determine whether the chosen software affects the results.

Software packages for deconvolution and statistical analysis were compared over a range of different samples to evaluate the workflow over a range of sample types; plant samples, solid human products (hair samples) and liquid human products (blood samples) were used. The different deconvolution software packages used showed different results through the studies showing that the software used will affect the outcome. Though some markers were consistent in the statistical analysis performed with the same deconvolution, a lot of the results were different which shows that conducting the analysis in different types of software, results in different biomarker detection. This could lead to a potential oversight and loss of important information.

The data showed that deconvolution worked best in Mass Profinder then statistical analysis in MPP gave the most reliable results whilst being the easiest to navigate. However, where possible, it was concluded that more than one type of software should be used for reliable biomarker detection to reduce the risk of losing important information through the software choice.

Citation

Scott, M. R. (2024). Evaluation of chemometric software for analysis of complex mixtures for biologically derived samples analysed using liquid-chromatography mass spectrometry. (Thesis). Keele University. Retrieved from https://keele-repository.worktribe.com/output/850250

• Thesis Type: Thesis
• Deposit Date: Jun 14, 2024
• Publicly Available Date: Jun 20, 2024
• Public URL: https://keele-repository.worktribe.com/output/850250
• Award Date: 2024-06