HiCcompare

HiCcompare performs joint normalization and differential analysis of Hi-C datasets to compare genome-wide chromatin interactions and identify condition-specific interaction changes, implemented as an R package on Bioconductor.

Key Features:

• Joint Normalization: Accounts for biases between multiple Hi-C datasets to enable direct comparative analyses across conditions such as disease states and distinct cell types.
• Loess-based Data-Driven Normalization: Employs locally weighted linear regression (loess) as a data-driven normalization strategy that avoids rigid parametric model constraints.
• Differential Analysis: Detects region-specific changes in chromatin interactions and identifies alterations that complement large-scale genomic rearrangements such as copy number variants (CNVs).
• Performance Benchmarking: Demonstrated superior performance relative to methods that normalize individual datasets in both simulated and real data contexts.

Scientific Applications:

• 3D Chromatin Architecture Studies: Comparative Hi-C analysis to investigate genome organization and its role in gene regulation.
• Tumor–Normal Comparisons: Identification of differential chromatin interactions between tumor and normal samples.
• Cell-Type-Specific Studies: Detection of interaction changes specific to distinct cell types.
• Disease Mechanisms and CNV Complementation: Exploration of chromatin interaction dynamics associated with complex diseases, including cancer, and integration with CNV analyses.

Methodology:

Processes chromosome-specific chromatin interaction matrices represented as three-column tab-separated text files in a sparse matrix format, performs joint normalization using locally weighted linear regression (loess), and conducts differential analysis to detect region-specific interaction changes while accounting for biases between datasets.