TY - JOUR ID - 10442/17668 A1 - Valavanis I. A1 - A1 - Sifakis E.G. A1 - A1 - Georgiadis P. A1 - A1 - Kyrtopoulos S. A1 - A1 - Chatziioannou A.A. Y1 - 2013/// T1 - Derivation of cancer related biomarkers from DNA methylation data from an epidemiological cohort JF - Communications in Computer and Information Science VL - 384 SN - 1865-0929 U3 - 10.1007/978-3-642-41016-1_27 PB - Springer Verlag SP - 249–256EP - UR - https://hdl.handle.net/10442/17668 N2 - DNA methylation profiling methods exploit microarray technologies and provide a wealth of high-volume data. This data solicits generic, analytical pipelines for the meaningful systems-level analysis and interpretation. In the current study, an intelligent framework is applied, encompassing epidemiological.DNA methylation data produced from the Illumina’s Infinium Human Methylation 450K Bead Chip platform, in an effort to correlate interesting methylation patterns with cancer predisposition and in particular breast cancer and B-cell lymphoma. Specifically, feature selection and classification are exploited in order to select the most reliable predictive cancer biomarkers, and assess their classification power for discriminating healthy versus cancer related classes. The selected features, which could represent predictive biomarkers for the two cancer types, attained high classification accuracies when imported to a series of classifiers. The results support the expediency of the methodology regarding its application in epidemiological studies. ER -