Résumé
This chapter describes a detailed account of the aforementioned modeling studies, giving ample space to the methodological challenges. It shows that models of the human chromosome 19 obtained by imposing the spatial proximity of loci known to be coregulated acquire large-scale features that are compatible with those of Hi-C map. The utilization of Hi-C data as constraints in three-dimensional genome modeling may require a conversion of chromosome-interaction frequencies to Euclidean distances. The zero-inflated negative binomial model is applied to each entry of Hi-C matrices, obtaining individual P-values for all possible pairwise interaction. As a prerequisite for the structural-functional analysis, the chapter examines to what extent the imposed proximity constraints, which are much fewer than the number of degrees of freedom of the model genome, suffice to pin down specific structural features. Genomic regions anchor to the nuclear lamina in large domains that are observed in single cells, but also display cell-to-cell variation.