A computational method to predict topologically associating domain boundaries combining histone Marks and sequence...

Area of Science:

• Genomics
• Computational Biology
• Epigenetics

Background:

• The three-dimensional (3D) chromatin structure is crucial for cell differentiation and development.
• Topologically associating domains (TADs) are conserved structural and functional units of chromatin architecture.
• Mechanisms underlying TAD boundary formation remain incompletely understood.

Purpose of the Study:

• To develop a computational method for inferring 3D chromatin structure, specifically predicting TAD boundaries.
• To investigate the role of epigenetic modifications and DNA sequence information in TAD boundary formation.

Main Methods:

• Development of TAD-Lactuca, a computational tool integrating epigenetic modification signals and primary DNA sequence data.
• Validation of TAD-Lactuca's stability across multiple resolutions and datasets.
• Comparative analysis of TAD-Lactuca against state-of-the-art methods.

Main Results:

• TAD-Lactuca demonstrates high accuracy in predicting TAD boundaries, outperforming existing methods when sequence patterns are included.
• The computational method is stable and reliable across different datasets and resolutions.
• Enrichment analysis revealed significant presence of transcription factor binding motifs, including non-CTCF motifs, at TAD boundaries.

Conclusions:

• TAD-Lactuca offers a cost-effective and efficient approach for predicting TAD boundaries.
• Incorporating sequence features substantially enhances prediction performance.
• The identified sequence motifs at boundaries support the role of TADs as functional units in gene regulation and highlight the importance of sequence patterns in chromatin folding.