Gladstone Institutes and the Arc Institute deployed AI to dismantle dogma that DNA sits locked inside nucleosomes. Long treated as tight spools barring entry, these structures instead keep over 85 percent of nucleosomes in partial distortion, with sections of DNA exposed rather than buried. The work rewrites how cells decide which genes to use and which to stash.

SAMOSA once mapped nucleosome positions, while the new IDLI tool scans internal architecture with machine learning to detect loose or missing protein blocks. Gladstone Institutes and Arc Institute teams traced 14 distinct nucleosome states tied to gene activity, showing transcription factors actively sculpt chromatin openness. Mouse stem cells and human-derived liver cells shared these patterns, confirming programmed control rather than random drift.

Vijay Ramani and Hani Goodarzi treat nucleosome distortion as a genomic volume dial that modulates gene output instead of a binary switch. Reading these states may clarify complex diseases and aging by exposing gradations in gene use. The researchers move from decoding chromatin grammar toward learning how to edit it.