VSS-Hi-C: variance-stabilized signals for chromatin contacts

Area of Science:

• Genomics
• Computational Biology
• Epigenetics

Background:

• Hi-C assays reveal 3D chromatin structure and function by measuring genomic locus interactions.
• Hi-C read counts exhibit heteroskedasticity, violating Gaussian assumptions and hindering analysis.
• Current variance stabilization methods are heuristic and cannot fully address mean-variance relationships.

Purpose of the Study:

• Introduce VSS-Hi-C, a data-driven method for variance stabilization of Hi-C data.
• Enhance visualization of Hi-C contact maps and improve downstream analyses.
• Provide a versatile R package for genomic and epigenomic data with replicates.

Main Methods:

• Developed VSS-Hi-C, a novel data-driven approach for variance stabilization.
• Applied VSS-Hi-C to Hi-C data to achieve unit variance signals.
• Validated performance improvements in visualization and subcompartment calling.

Main Results:

• VSS-Hi-C signals demonstrate unit variance, significantly improving Hi-C heatmap visualization.
• The method enhances the performance of computational tools relying on Gaussian assumptions.
• VSS-Hi-C is available as an R package for broad applicability.

Conclusions:

• VSS-Hi-C offers a superior data-driven solution for variance stabilization in Hi-C data.
• The method facilitates more accurate visualization and analysis of 3D genome organization.
• VSS-Hi-C is a valuable tool for genomic and epigenomic research with replicates.