Author/Editor		Slavkov, Ivica; Džeroski, Sašo; Peterlin, Borut; Lovrečić, Luca
Title		Analysis of gene expression profiles for Huntington's disease patients with predictive clustering trees
Type		članek
Source		In: Zdravje na informacijski poti. Zbornik kongresa Slovenskega društva za medicinsko informatiko; 2006 apr 9-11; Zreče. Ljubljana: Slovensko društvo za medicinsko informatiko,
Publication year		2006
Volume		str. 164-75
Language		eng
Abstract		ln this paper we analyzed microarray data of patients with Huntington's disease (HD). A common problem in microarray data analysis is a small sample size that can be obtained for a study of a certain disease. This problem is sometimes worsened because expression levels from microarray data obtained from different runs are not directly comparable to each other. In this paper the expression levels of ribosomal protein genes were used for two different tasks. First, as a way to test if re-normalization between different sets (runs) of HD microarray data is necessary and second, they were used to re-normalize the data. Ribosornal genes were chosen for this task because they have relatively stable expression levels across the runs and their expression is not dependent on the patient/control status. After this preprocessing step of the microarray data, Predictive Clustering Trees (PCTs) were used to identify useful gene expression profiles and also to connect patient records with gene expression levels. The PCTs were constructed by using a generic system for constructing decision trees. ln the end patients' pathological characteristics which created the biggest difference in gene expression levels were identified, but also genes that could possibly serve as a way of differentiating between patients and control subjects, or presymptomatic and symptomatic patients.
Descriptors		HUNTINGTON'S DISEASE GENE EXPRESSION OLIGONUCLEOTIDE PROBES DECISION TREES RIBOSOMES CLUSTER ANALYSIS