Nucleosome-resolution inference of chromatin interaction landscapes from Micro-C data using maximum entropy modeling

Inferring the physical interaction landscape underlying chromatin contact maps remains a central challenge in genome biology. Chromosome conformation capture experiments such as Micro-C provide high-resolution measurements of spatial contacts between genomic loci, yet translating these contact frequencies into an interpretable structural interaction model is a fundamentally underdetermined inverse problem. Here we present a nucleosome-resolution maximum entropy framework that infers effective pairwise interaction parameters directly from experimental Micro-C data. In this approach, chromatin is represented as a heterogeneous nucleosome-linker polymer, and maximum entropy optimization identifies the minimal set of interaction constraints required to reproduce the observed contact statistics. Applied to several human genomic loci, the inferred interaction landscape accurately reconstructs experimental contact maps and remains robust to substantial masking or perturbation of the input data. Forward simulations using the inferred parameters generate structural ensembles that reproduce the observed chromatin organization, demonstrating that the model captures the generative constraints governing chromatin folding rather than merely fitting contact frequencies. Analysis of these ensembles reveals domain boundaries, localized interaction hotspots and emergent chromatin "blobs" that coincide with independently observed architectural features. Together, these results establish a highresolution statistical inference framework that reconstructs locus-scale chromatin structure while providing a physically interpretable interaction landscape linking chromatin contact maps to three-dimensional genome organization.