A quantile approach to analyzing differential methylation hybridization (DMH) microarrays
Mathematical Biosciences Institute (MBI), The Ohio State University
(January 15, 2009 2:30 PM - 3:30 PM)
DNA methylation has been shown to play an important role in the silencing of tumor suppressor genes in various tumor types. As it is desirable to have a system-wide understanding of the methylation changes that occur in tumors, we have developed the DMH protocol that can simultaneously assay the methylation status of all known CpG islands. As is common with all microarray technologies, a large percentage of the signal obtained from the array can be attributed to various measurable and unmeasurable confounding factors unrelated to the biological question at hand. In order to correct the bias due to the noise, we have implemented a quantile regression model for assessing significance of signal. As a proof of concept, we have applied this model to the methylation signature analysis of breast cancer cell lines which were obtained from the LBNL-ICBP (Lawrance Berkeley National Lab, Integrative Cancer Biology Program). We have been able to correctly identify some known methylated and unmethylated genomic regions. In this paper, we will describe how to use quantile regression to model DNA methylation microarrays and discuss how to summarize the regression results to identify all regions that are commonly methylated among all breast cancer cell lines. Finally, we validate our results using known methylated genes and housekeeping genes.