Institute of Information Science, Academia Sinica

Genome sequence alone is insufficient to account for the diversity of cell types and functions within a single individual, nor for the phenotypic and disease diversity across individuals. These are in part captured by epigenetic changes in the chemical modifications of the DNA and its packaging, which are important indicators, and sometimes important actors, of gene-regulatory mechanisms. In this talk, I will present computational, statistical, and data science approaches for epigenetic data analysis at the genome-wide scale. First, I discuss an epigenomic signature for olfactory receptor genes that enables monogenic and monoallelic expression. Second, I discuss integration of DNA methylation data with chromatin state profiles and cell types to reveal epigenetic changes during differentiation.  Third, I discuss ChromDiff, a novel statistical and information-theoretic computational methodology to identify chromatin state differences in gene bodies and linked regulatory regions across groups of samples. Lastly, I discuss application of ChromDiff to characterize signatures of sex, developmental age, and tissue type, unveiling both the chromatin states and the gene classes that distinguish each group of epigenomes. Overall, the methods presented are general, and should be applicable to a broad range of epigenomic analyses.