Analysis of SELDI mass spectra for biomarker discovery and cancer classification

Cheng, Yaping (2009). Analysis of SELDI mass spectra for biomarker discovery and cancer classification. University of Birmingham. Ph.D.

Preview

Cheng09Phd.pdf
PDF
Download (1MB)

Abstract

The thesis focused on data analysis of surface-enhanced laser desorption/ionisation time-of-flight mass spectrometry (SELDI-TOF MS) for biomarker discovery and cancer classification. It investigated quantitative measures of reproducibility and found that SELDI protein profiles are affected by sample storage and processing procedure. Two new peak alignment algorithms were proposed, one of which achieved the best performance when compared to the existing methods. The assumption of normality of SELDI protein profiles, on which the standard statistical methods are based, was examined. Normality tests and the multiple testing procedures revealed that SELDI protein profiles do not follow normal distributions, implying that it may be reliable to use non-parametric methods for detecting disease-associated proteins. A new normalisation algorithm was proposed, and was shown to give a better improvement of normality compared with the existing methods. An integrated algorithm to discover proteomic biomarkers for cancer diagnosis was proposed and applied to two published SELDI data sets. The results demonstrated that the receiver operating characteristic (ROC) curve method may be more reliable to determine the discriminatory powers of the identified biomarkers compared to Wilcoxon test. The methods for proteomic biomarker discovery presented here may be generalisable and applicable to other mass spectrometry and genomics approaches.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):