Identification of differential topologically associating domains from low sequencing depth and pseudobulk chromatin contact maps.

HiDT accurately detects differential topologically associating domains with high precision at low sequencing depths, outperforming existing methods across various datasets.

• Why it matters: Understanding TAD reorganization is crucial for insights into development, disease, and gene regulation, but current approaches struggle with low-depth or pseudo-bulk data, limiting discovery.
• What they did: The authors developed HiDT, a graph neural network-based algorithm with an attention-enhanced layer, trained on diverse sequencing depths to identify differential TADs reliably.
• The result: HiDT enables robust detection of TAD changes linked to oncogene dysregulation, transcriptional heterogeneity, and structural variations, advancing chromatin architecture analysis in challenging datasets.