TADBD: a sensitive and fast method for detection of typologically associated domain boundaries

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Topologically Associated Domains (TADs) are crucial genomic structures influencing gene regulation.
• Accurate detection of TAD boundaries from Hi-C contact matrices is essential for understanding 3D genome architecture.
• Existing methods for TAD boundary detection face limitations in speed and accuracy.

Purpose of the Study:

• To present TAD boundary detection (TADBD), a novel computational method for sensitive and rapid identification of TAD boundaries.
• To provide an efficient tool for analyzing 3D genome organization at the TAD level.

Main Methods:

• TADBD utilizes a Haar-based algorithm incorporating a Haar diagonal template.
• The method employs acceleration through a compact integrogram and multi-scale aggregation.
• Statistical filtering is applied for robust boundary identification.

Main Results:

• TADBD demonstrated high sensitivity and speed in detecting TAD boundaries on Hi-C contact matrices.
• Comparative analyses using simulated and experimental data showed TADBD outperformed five other existing methods.
• The method's effectiveness was validated across diverse datasets.

Conclusions:

• TADBD offers a significant advancement in the computational analysis of 3D genome architecture.
• The developed R package for TADBD is freely available, promoting wider adoption and research.
• This tool facilitates more accurate and efficient studies of TADs and their functional implications.