Abstract:
|
Cytosine methylation is increasingly being used as a diagnostic and prognostic biomarker. Currently, this is largely limited to detecting regions of aberrant hypermethylation. However, because methylation is a semi-stable epigenetic modification, governed by a complex set of dynamic processes, its use as a biomarker could be improved by developing techniques to detect not only abnormal levels of methylation but also abnormal methylation dynamics. Early stochastic models of methylation pattern inheritance, conceived in the absence of detailed biological data, fail to capture the rich spatial dependence which has been observed in nature and are unable to replicate the types of patterns seen in biological samples and are so unsuitable for detecting unusual dynamics. Here, we develop and test a class of novel, distance-dependent Markov models which are able to replicate biological structures. The models are then used to profile various test tissue samples, and their potential use as prognostic or diagnostic tools is discussed.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.