Cross-Scale Convergence in Epigenetic Gene Regulation: A Perspective on Functional Enrichment Analytics for Cancer.

Epigenetic regulation of gene expression is studied at three physical scales: micro: DNA sequence-level methylation/demethylation; meso: nucleosome occupancy and remodeling; and macro: chromosomal domain silencing by Polycomb complexes, heterochromatin, and topologically associating domain (TAD) boundaries. The challenge to fully understand epigenetic gene regulation patterns is that these scales are not independent. Their influence overlaps and they share a recurring architectural theme across scales of a targeted molecular pattern followed by cooperative, feedback-driven, spatially bounded spreading. We argue here that disruption of this shared architecture at any one scale is independently sufficient to tip a bistable silencing domain into an oncogenic state. This paper discusses how such a cross-scale architectural rule set has concrete implications (yet underexploited) for computational cancer epigenomics, e.g., functional enrichment analyses generally focus on epigenetic features at one scale as an independent line of evidence, ignoring corroborating signals that could be reinforced by underlying hierarchical levels. This paper outlines options for functional enrichment statistics that combine multiple corroborating molecular features within a scale and corroborating evidence across scales into composite confidence scores calibrated against an empirical null that preserves correlations between assays. We propose benchmarking this approach against conventional single-feature enrichment in matched multi-omic cancer datasets as a direct test of the model.