Computational Epigenomics: From Fundamental Research to Disease Prediction and Risk Assessment

ABSTRACT

Over the past two decades, rapid advances in DNA sequencing technologies have allowed genome-wide interrogation of epigenetic features. The epigenome landscape encompasses a growing number of chemical properties of DNA and DNA-associated proteins; these properties are tissue-specific, distinctive for disease state and sensitive to environmental exposures. The epigenetic field has rapidly evolved from basic research investigations, aiming to understand the nature and function of epigenetic marks, to clinical and preclinical applications, where vast epigenetic information is used for risk assessment and disease prediction. The large diversity of epigenetic marks is mirrored by the complex variability of their genomic patterns and distributions. Mining of large-scale genomic datasets relies strongly on computational approaches and statistical models that should be carefully selected and adapted to fit the nature of the signals analyzed and the hypotheses tested. Here, we review recent advances in computational approaches used to analyze epigenetic data, with an emphasis on histone modifications and DNA methylation. We discuss the standard workflows for data acquisition, processing, and transformation, as well as the computational approaches used to assess statistical significance in comparative analyses. We also discuss the prediction methods utilized to associate epigenetic modifications with human disorders and environmental factors.