A technique for preserving network structure in randomized Hi-C data

Chromatin interaction data are frequently analyzed as a network to study several aspects of chromatin structure. Hi-C experiments are costly and there is a need to create simulated networks for quality assessment or result validation purposes. Existing tools do not maintain network properties during randomization. We propose an algorithm to modify an existing chromatin interaction graph while preserving the graphs most basic topological features - node degrees and interaction length distribution. The algorithm is implemented in Python and its open-source code as well as the data to reproduce the results are available on Github.