miniMDS: 3D structural inference from high-resolution Hi-C data.

Area of Science:

• Genomics
• Computational Biology
• Structural Biology

Background:

• High-throughput chromosome conformation capture (Hi-C) experiments generate valuable data for understanding 3D genome organization.
• Existing computational methods struggle with the scale and resolution of modern Hi-C datasets, limiting 3D structural inference.
• Inferring the three-dimensional structure of the human genome at high resolution (10 kbp) presents significant computational challenges.

Purpose of the Study:

• To develop a computationally feasible method for 3D genome structural inference from high-resolution Hi-C data.
• To improve the speed, accuracy, and memory efficiency of 3D genome modeling compared to existing techniques.

Main Methods:

• Developed miniMDS, an approximation of multidimensional scaling (MDS).
• miniMDS partitions Hi-C datasets for parallel, high-resolution MDS analysis.
• Reassembles partitioned data using low-resolution MDS for efficient genome-wide reconstruction.

Main Results:

• miniMDS significantly outperforms existing methods in speed, accuracy, and memory usage for 10 kbp human genome inference.
• The method enables feasible 3D structural inference from previously computationally intractable high-resolution Hi-C datasets.
• Demonstrated the effectiveness of the partitioning and reassembly strategy for large-scale genomic data.

Conclusions:

• miniMDS provides a scalable and efficient solution for 3D genome structure inference.
• The developed method facilitates deeper insights into genome organization and function using high-resolution Hi-C data.
• A Python implementation of miniMDS is publicly available for the research community.