Fig. 1: An updated mouse embryonic stem cell chromatin contact map produced with multiplex-GAM.

a, In a standard GAM experiment, thin slices from individual nuclei (nuclear profiles, NPs) are isolated by cryosectioning and laser microdissection, before the DNA content of each slice is determined by next-generation sequencing. In a multiplex-GAM experiment, DNA from multiple NPs is extracted and sequenced together, reducing sequencing costs. b, Multiplex-GAM data constructed in silico by combining 1NP datasets at 40 kb resolution (chromosome (chr.) 6, 49–54 Mb). Contact maps were produced from single-NP data (top), in silico 2NP data (middle) and in silico 3NP data (bottom). Dʹ, normalized linkage disequilibrium. c, The updated SLICE model accounts for the number of NPs multiplexed in each tube (X_NP), the nuclear ellipticity (ε) and the thickness of each NP (h). d, SLICE models can be used to guide the experimental design, for example to estimate the minimum number of tubes (m*) needed to achieve a given statistical power. e,f, Visualization of contacts centered at the Sox2 locus (chr. 3, 30–39 Mb) in GAM 1NP data and 3NP data (e), and the combined GAM-1,250 dataset (f), all at 40 kb resolution. g, Significant pairwise interactions at 40 kb resolution identified by SLICE between functional elements in the Sox2 locus, including the Sox2 gene and its closest super-enhancer (SE). The arrows indicate previously identified interactions between these elements³⁸. Neur. E, neuronal enhancer. h,i, Enrichment analysis of pairwise interactions (h) and triplet interactions (i) identified by SLICE involving active, inactive, intergenic or enhancer regions (h) and topologically associating domains that are highly transcribed (high), lowly transcribed (low) or that overlap super-enhancers. (i) for the GAM-1,250 dataset and the original GAM-408 dataset. Statistically significant enrichments or depletions (those falling outside 95% of randomized observations after Bonferroni correction) are marked by an asterisk.