Background: The systematic identification of transcriptional repressors remains challenging, as current inference frameworks are predominantly optimized for accessible chromatin, leaving regulatory signals embedded within repressive domains undercharacterized. Methods: Here, we present NuRepress, a computational framework that predicts candidate transcriptional repressors by integrating repressive chromatin architecture, functional signatures, and transcriptional outcomes. NuRepress first identifies well-phased nucleosome arrays within repressive chromatin. These arrays are treated as discrete structural units that capture characteristic local chromatin organization associated with regulatory activity. Since distinct Tn5 cut signal patterns often imply divergent regulatory functions, the framework stratifies these arrays into potential functional subtypes. By synthesizing the quantified repressive efficacy of each subtype with spatial motif enrichment and observed transcriptional dynamics, NuRepress systematically prioritizes and ranks candidate repressors. Results: Our analysis indicated that well-phased nucleosome arrays exhibited accessibility-defined organizational patterns with distinct repressive efficacies, and that these patterns were also observed across species, suggesting that the structural principles captured by NuRepress might extend beyond one specific biological system. Positional motif analysis revealed that distinct TFs exhibited different spatial preferences relative to well-phased nucleosome arrays, suggesting scale-specific preferences for their interactions with these organized chromatin structures. When applied to pancreatic cancer progression, NuRepress identified changes in nucleosome organization associated with stage-specific transcriptional remodeling, highlighting candidate repressors of key oncogenic drivers. Conclusions: NuRepress establishes a structure-aware strategy for repressor inference that extends regulatory genomics beyond accessibility-centered paradigms. By linking well-phased nucleosome organization to transcriptional outcomes, it provides a principled framework for dissecting transcriptional repression across diverse biological settings.
Xiang et al. (Sat,) studied this question.