Analyzing Microarray Data Using Homogenous and Inhomogenous Hidden Markov Models
Master's Thesis, Martin-Luther-Universität Halle, 2006.
Chromosomal imbalances and gene expression alterations play a central role in different types of cancer. Microarray experiments are common techniques to generate data sets which contain such information. This diploma thesis has two goals. The first goal is to develop inhomogeneous Hidden Markov Models to detect chromosomal imbalances and gene expression alterations and the second goal is to test the performance of this novel approach on breast cancer data. To achieve these goals we have extended the mathematical theory of standard Hidden Markov Models to obtain inhomogeneous Hidden Markov Models with a few easily interpretable parameters. The improved quality of our novel approach in comparison with the standard Hidden Markov Models is the result of using the chromosomal locations of genes and the microarray measurements of theses genes as input data for our Hidden Markov Models. To our knowledge the simultaneous usage of both, the chromosomal locations and the microarray measurements, is a novel strategy. Previously described methods use the chromosomal locations only to interpolate the data. The fact that our inhomogeneous Hidden Markov Models are able to find known candidate genes for over-expression and in literature described losses and gains of DNA segments in breast cancer data represents the good quality of this approach.
A reprint is available as PDF.
The publication includes results from the following projects or software tools: ArrayCGH.
Further publications by Michael Seifert.