Apparent 3D-structural variant-effect signal is explained by variant category, not structure: a category-matched evaluation across nine disease loci
This study demonstrates that the apparent predictive power of a 3D chromatin structural-disruption score (ARCHCODE) for pathogenicity is largely an artifact of variant category rather than genuine structural information, as its performance collapses to chance levels when evaluated using category-matched controls, unlike established predictors such as CADD and phyloP.