Bio: Epigenetics

This working group will focus on two problems in epigenetics, one on DNA methylation and the other on three-dimensional chromosomal organization.

DNA Methylation: The cornerstone of epigenomics is methylation. With NGS, nucleotide resolution methylation (NTRM) data are increasingly available, but such data present challenges because of the following features: (1) measured signals may cover multiple sites; (2) methylation intensities in different sites are typically highly correlated; (3) there are undetectable signals in many sites within a region of interest; (4) sample sizes are typically small, which leads to the question of how to pool information from multiple regions. To date, a common approach is to average signals across nucleotides within a region, which ignores many of the features, not the least is the correlation between the signals. Thus far, appropriate statistical tools that take the special features of NTRM data into consideration are lacking. Although GEE has also been used to account for correlation, but the results are less than completely satisfactory due to the other unusual features.

3-D Regulation: The expression of a gene is usually believed to be controlled by the regulatory elements bind to its promoter region. However, it has long been observed that, in complex genomes, such as the human genome, a gene may be controlled by distal enhancers and repressors; some may even be located on different chromosome than the genes that they regulate. A recent molecular technique, 3C (chromosome conformation capture), which uses formaldehyde cross-linking and locus-specific PCR, is able to detect physical contacts between distant genomic loci. Such communication is achieved through spatial organization (looping) of chromosomes to bring genes and their regulatory elements into close proximity. Several adaptations of the 3C assay to study genomewide spatial interactions, including Hi-C, have been developed. The availability of such data makes it possible to reconstruct the underlying three-dimensional spatial chromatin structure and to study their roles in disease development. The challenging statistical issues include: (1) the experimental procedure for inducing spatial interactions may lead to "random collisions" (false loops) that need to be filtered out; (2) the data are highly correlated due to many-to-many mappings because one gene may be regulated by multiple sites and a regulator may controls many genes; (3) very large dimensionality of the parameter space for reconstructing 3D structures.