CSSIG

Abstract

The research on the causes of cancer and its precision medicine has been the focus of scientists. DNA methylation, as an important epigenetic modification, has high heterogeneity among samples, which has been shown to be important in the development and progression of cancer. Therefore, identification of sample-specific pathogenic genes based on DNA methylation is one of the most fundamental and key issues in achieving precision medicine from the perspective of single sample. Considering the high heterogeneity of DNA methylation, we propose an approach, CSSIG, to identify cancer sample-specific genes through extracting the comprehensive features of DNA methylation data using PCA and obtaining a set of sample-specific genes by gain model based on information entropy theory. Application to the real breast cancer data set, 31 cancer significant sample-specific genes are identified. Through analyzing these gene, it is shown that the biological significance of these genes in samples, and thus understand the different mechanisms in disease in each sample.