HiCDB: a sensitive and robust method for detecting contact domain boundaries

Nucleic Acids Res. 2018 Nov 30;46(21):11239-11250. doi: 10.1093/nar/gky789.

Abstract

Contact domains are closely linked to gene regulation and lineage commitment, while current understanding of contact domains and their boundaries is still limited. Here, we present a novel method HiCDB, which is constructively based on local relative insulation metric and multi-scale aggregation approach to detect contact domain boundaries (CDBs) on Hi-C maps. Compared with other 'state-of-art' methods, HiCDB shows improved sensitivity and specificity in determining CDBs at various Hi-C resolutions. The superiority of HiCDB enabled us to study the epigenetic features of detected CDBs and showed enrichment of architectural proteins and cell-type-specific transcription factor binding sites at CDBs. The further comparison of GM12878 and IMR90 Hi-C datasets suggested that cell-type-specific CDBs are marked by active regulatory signals and correlate with activation of nearby cell identity genes.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Algorithms*
  • Animals
  • Binding Sites
  • Cell Line
  • Computational Biology / methods*
  • Enhancer Elements, Genetic / genetics
  • Gene Expression Profiling*
  • Gene Expression Regulation*
  • Humans
  • Insulator Elements / genetics
  • Mice
  • Promoter Regions, Genetic / genetics
  • Protein Binding
  • Protein Domains*
  • Reproducibility of Results