Fig. 1

Overview of the population-based genome structure modeling approach and its application to the Drosophila genome. a The initial structures are random configurations. Maximum likelihood optimization is achieved through an iterative process with two steps, assignment (A) and modeling (M). We increase the optimization hardness over several stages by including contacts from the Hi-C matrix A with lower probability thresholds (θ). After the population reproduces the complete Hi-C data, we include the vector E (lamina-DamID), again in stages with decreasing contact probability thresholds (λ). b Schematic of the Drosophila genome. The autosome arms are designated 2L, 2R, 3L, 3R, 4, and X. The arms of chr2 and chr3 are connected by centromeres labeled “C”. Euchromatic regions are labeled as the arm. The numbers along the top of a genome indicate the length of the section in megabases (Mb), and for euchromatin the number of spheres (TADs) in the structure model is also given. The heterochromatic region of each chromosome arm is labeled “H”. The white gene is located ~19 M away from the heterochromatin of chrX. Also indicated are the Hox genes: five genes of the Antennapedia complex (ANT-C) are located at ~2.3–2.8 Mb from the heterochromatin of chr3R, and three genes of the Bithorax complex (BX-C) are located at ~12.4–12.7 Mb from the heterochromatin of chr3R. c Snapshot of a single structure randomly picked from the final population. Left panel: The full diploid chromosomes are shown in colors: blue, chr2; green, chr3; magenta, chr4; orange, chrX. The two homologs of the same chromosome are distinguished by the color tone, with one homolog copy with lighter and one with darker color. The heterochromatin spheres are larger than the euchromatin domains. The nucleolus is colored in silver. Right panel: The euchromatin domains are colored to reflect their epigenetic class: red, active; blue, PcG; green, HP1; dark purple, null. Heterochromatin spheres are shown in grey and the nucleolus in pink