3D Genome Reconstruction from Partially Phased Hi-C Data

Area of Science:

• Genomics and Bioinformatics
• Computational Biology
• Structural Biology

Background:

• The three-dimensional (3D) genome structure is crucial for cellular functions.
• Reconstructing 3D chromosome structures from Hi-C data in diploid organisms presents unique challenges compared to haploid organisms.

Purpose of the Study:

• To investigate the identifiability of 3D chromosome structures from Hi-C data in diploid organisms.
• To develop a novel computational method for 3D genome reconstruction using algebraic geometry and semidefinite programming.

Main Methods:

• Application of algebraic geometry techniques to analyze the identifiability of 3D genome structures.
• Development of a 3D reconstruction method combining semidefinite programming, numerical algebraic geometry, and local optimization.
• Validation using simulated datasets with varying noise levels and amounts of phased data.

Main Results:

• Proof that a small quantity of phased Hi-C data is sufficient for finite identifiability of 3D genome structure, even with noise.
• Demonstration of the proposed method's performance on simulated data, showing robustness across different noise levels.
• Successful application to a real mouse X chromosome dataset, recovering known structural features.

Conclusions:

• The study establishes theoretical grounds for 3D genome structure reconstruction from limited phased Hi-C data in diploid organisms.
• The proposed computational method offers a robust and effective approach for accurate 3D genome structure recovery.