Linking Hi-C and Spectroscopic Single-molecule Socalization Microscopy (sSMLM) Experiments through Molecular Simulation and Machine Learning

Super-resolution optical microscopy techniques such as spectroscopic single-molecule localization microscopy (sSMLM) have shown great potential to enable structural analysis of chromatin across the genome with nanometer-scale resolution and time-domain information. However, progress remains to be made in analysis techniques which allow the results of such microscopy methods to be compared with complementary non-microscopic techniques such as Hi-C, Chip-seq, and ATAC-seq. In this work, we introduce analytical formulas for a descriptive quantity, the epigenetic molar contact ratio, which characterizes the ‘plaid pattern’ detectable in Hi-C maps, and which we show can be calculated directly from both Hi-C data and the results of well-established sSMLM techniques. We validate this technique by comparison to a ground-truth polymer model, both determining the performance of this model and identifying the biological systems in which we expect this technique to be more or less accurate. We also demonstrate the ability to quantitatively reconstruct the Hi-C map of a polymer by performing this analysis on in silico-generated sSMLM images. Our analysis demonstrates the potential for advanced analytical techniques to predict complex parameters from individual sSMLM images in single cells.