Bladder Cancer Prediction Using Genetic Algorithm and Fuzzy Rule-Based System

The aberrations in the DNA Methylation levels are known cause diseases like Cancer by regulating the gene expression. The effect of DNA Methylation in Bladder Cancer is studied in this paper by integrating the gene expression data and DNA methylation data. A pipeline is proposed to select the biomarker genes for the disease and to use them in the classification of samples into "Normal" and "Cancer" classes. The pipeline comprises differential analysis of DNA methylation and gene expression datasets in "Normal" and "Cancer" samples, overlapping the results with epigenetically relevant genes in Epifactors database, and applying Genetic Algorithm for the selection of optimal feature set. Then, a Fuzzy Rule-Based System is modelled for cancer prediction which obtained an accuracy of 100%. The biological significance of the biomarkers is verified by checking against the KEGG database. The selection of the most informative genes for Bladder Cancer and production of interpretable rules are the major contributions of this work. The method can be extended to pathways in Bladder Cancer.