Regulatory-scale stripe analysis from single-cell Hi-C with scStripe

Abstract

Chromatin stripes are a directional architectural feature of three-dimensional genome organization, but their analysis in single-cell Hi-C (scHi-C) at regulatory resolution is hindered by extreme sparsity. We developed scStripe, an imputation-free two-stage statistical framework that detects stripes from sparse cell-type--aggregated scHi-C maps and directly quantifies the resulting aggregate-defined stripes in raw individual-cell contact maps using per-cell stripe scores. Across three complementary benchmark settings, scStripe achieved the highest F1 scores and showed strong aggregate stripe enrichment, while the detected stripes recapitulated characteristic structural and regulatory features reported in bulk data. Per-cell stripe scores resolved recurrent configurations of the EBF1-anchored stripe, captured genome-wide variation associated with cell type, cell-cycle phase, and developmental stage, and identified positively associated gene--stripe pairs with pronounced cell-type specificity using matched scRNA-seq. Together, scStripe provides a practical, scalable framework for regulatory-scale stripe analysis across single cells without requiring single-cell enhancement or 3D reconstruction.