MrTADFinder: A network modularity based approach to identify topologically associating domains in multiple...

Area of Science:

• Genomics and Bioinformatics
• Network Science
• Computational Biology

Background:

• Eukaryotic genomes are organized into topologically associating domains (TADs), revealed by Hi-C assays.
• TAD organization is complex, exhibiting varied length scales and nested structures.
• Identifying TADs and their multi-scale organization requires advanced computational methods.

Purpose of the Study:

• To develop a novel computational framework, MrTADFinder, for identifying TADs from Hi-C data.
• To explore the multi-scale organization of TADs and their relationship with genomic features.
• To analyze the role of chromatin marks, transcription factors, and somatic mutations at TAD boundaries.

Main Methods:

• Formulated TAD identification as a network optimization problem using the concept of modularity.
• Developed a background model for random chain contacts by solving matrix equations, preserving coverage and distance dependence.
• Introduced a tunable resolution parameter for multi-resolution TAD identification.

Main Results:

• MrTADFinder successfully identified TADs in Hi-C datasets, with boundaries showing characteristic chromatin mark and transcription factor signatures.
• Multi-resolution analysis revealed resolution-dependent changes in boundary signatures and identified characteristic length scales for different chromatin features.
• Observed enrichment of high-occupancy target regions near TAD boundaries and a stepwise pattern of somatic mutations across boundaries.

Conclusions:

• MrTADFinder provides a robust computational framework for dissecting multi-scale genomic structures in Hi-C data.
• The study highlights the intricate relationship between TAD organization, epigenetic regulation, and genomic alterations.
• The multi-resolution approach offers new insights into the dynamic nature of genome architecture.