Cancer is both an epigenetic and genetic disease. Ovarian cancer is one of the most dangerous cancers among other cancers of the female. TCGA has provided a comprehensive of epigenomics and genomics data for the biomedical research community. This data can be used to analyze genomic alteration in very large scale to understand how certain disease may develop.
This study retrieved from the TCGA database a total of 12 batches of DNA methylation and microRNA (miRNA) for the ovarian cancer. The method of K-Nearest Neighbors imputation was used to deal with the missing data problems. The problem of data heterogeneity, can be solved by using the meta-analysis approach.
Three different cutoffs were introduced to rank the results of DNA-methylated miRNA events. These miRNAs were compared with experimental databases to validate our findings. We obtained 31 miRNAs that are potentially epigenetic-mediated and involved in ovarian cancer formation. In particular, at least five miRNA findings (hsa-miR-150, hsa-miR-335*, hsa-let-7f-1*, hsa-miR-197, hsa-miR-140-3p) has been confirmed to be involved in ovarian cancer, although no previous study addresses the epigenetic mechanism.
In conclusion, our analysis pipeline provided a systematic method to identify potential miRNA candidates which can possibly be silenced or activated by epigenetic mechanisms in ovarian cancer. The use of imputation and meta-analysis allow us to construct highly confident DNA-methylation-mediated miRNAs as prognostics biomarkers for cancer studies.
