Chromatin Architecture Reconstruction

Abstract

Three-dimensional genomic architecture is increasingly being found to play an important role in the regulation of gene expression through direct physical interactions of distant genomic sites. Conformation Capture Techniques (CCTs) attempt to resolve the conformation of a genetic sequence or entire genome by measuring mean interaction frequencies between pairs of genomic loci in a population of fixed cells. Essential to the interpretation of these measurements is the use of a molecular model to infer genomic conformation from measured interaction frequencies. Unlike macromolecules that exhibit a single, dominant conformation in their native state, chromatin exhibits conformational polymorphism at multiple scales that renders optimization techniques that solve for a single, dominant conformation in the structure inference process suboptimal for the interpretation of CCT data. As an alternative, we present a computational procedure that solves for the unique, maximum entropy structural ensemble that is consistent with experimentally measured interaction frequencies. By modeling genomic topology and treating the full conformational ensemble in the fitting process, ensemble average structural quantities including correlations in distances between distinct pairs of genomic sites are obtained in addition to mean interaction frequencies between all genomic sites. Application of the procedure to the human HoxA cluster suggests that it is organized into multiple chromatin loops in differentiated cells. The present approach that is founded on the principle of maximum entropy is equally applicable to fluorescence-based data as obtained, for example, from fluorescence in situ hybridization as it is to CCT-based measurements.