Dice Question Streamline Icon: https://streamlinehq.com

Develop methods to predict differential 3D chromatin structures and identify their driving genomic features

Develop machine learning methods to predict differential three-dimensional chromatin changes between conditions—specifically gained, lost, and conserved topologically associating domain (TAD) and loop boundaries, as well as differential A/B compartments and chromatin stripes—and ascertain the genomic features associated with these differential structures.

Information Square Streamline Icon: https://streamlinehq.com

Background

The paper emphasizes that 3D genome organization varies across cell types, cell-cycle stages, and conditions, with a significant fraction of TADs being facultative. While statistical and overlap-based analyses can detect differential structures, predictive methods that learn features driving such changes are scarce.

A principled predictive framework would enable discovery of genomic determinants of structural remodeling (e.g., factors and histone marks) and facilitate interpretation of 3D changes in development and disease.

References

Despite the biological relevance and interpretability of such 3D changes, methods for predicting them, and consequently the knowledge of associated genomic features, remain undeveloped and a highly promising direction for future investigation.

— Machine and deep learning methods for predicting 3D genome organization (2403.03231 - Wall et al., 4 Mar 2024) in Discussion, paragraph on differential 3D structures and their prediction