Towards a robust algorithm to determine topological domains from colocalization data

摘要

One of the most important tasks in understanding the complex spatialorganization of the genome consists in extracting information about thisspatial organization, the function and structure of chromatin topologicaldomains from existing experimental data, in particular, from genomecolocalization (Hi-C) matrices. Here we present an algorithm allowing to revealthe underlying hierarchical domain structure of a polymer conformation fromanalyzing the modularity of colocalization matrices. We also test thisalgorithm on several model polymer structures: equilibrium globules, randomfractal globules and regular fractal (Peano) conformations. We define what wecall a spectrum of cluster borders, and show that these spectra behavestrikingly differently for equilibrium and fractal conformations, allowing usto suggest an additional criterion to identify fractal polymer conformations.