Cancer Specific Epigenetic Biomarker Extraction from DNA Methylation and Gene Expression Data

The epigenetic aberrations are known to be involved in the process of tumorigenesis by the overexpression of oncogenes and underexpression of tumor suppressor genes. This mechanism can be explained by identifying the genes that are up or down regulated due to epigenetic activities. Here, DNA methylation, an important epigenetic activity is considered for extracting the epigenetically significant biomarker genes from 12 types of cancers. For each cancer type, a differential analysis is performed on gene expression and DNA methylation separately and the resultant gene sets are overlapped to generate the epigenetically important statistically significant genes of that cancer as biomarkers that have different distribution of values on "Cancer" and "Normal" samples. The identified genes are found to be involved in different carcinogenetic and related biological pathways and can be utilized for personalized treatment strategies. The disease-specific genes achieved an accuracy of 96.19% in the cancer classification task using a random forest model.