Proceeding from Hi-C contact level data to inferred 3D genomic configurations confers biological advantages. Mitigating these are concerns surrounding the accuracy of reconstructions given the absence of gold standards, competing reconstruction algorithms, and their tuning sensitivities. Previously, 3D reconstruction accuracy was addressed by recourse to a limited number (~5-6) of low resolution FISH probes: the handful of inter-probe distances being correlated with corresponding reconstruction-derived distances to provide a crude, unstable concordance measure.
New multiplexed FISH (mFISH) and proximity ligation-free genomic architecture mapping (GAM) assays enable rigorous approaches to 3D genome reconstruction accuracy assessment. Using a state-of-the-art reconstruction algorithm, HSA, we obtain 3D configurations of chromosomes 20, 21 and 22 using Hi-C data from IMR90 cells and reference these against mFISH imaging. Crucially, the availability of mFISH replicates permits formal inference.
GAM is based on ultrathin cryosectioning enabling validation by assessing agreement with underlying planar slices. We apply this approach to mESCs, utilizing permutation-based inference.
|