Maass' Lab Develops Machine-Learning Algorithm for Inter-chromosomal Mapping

Toronto, Ontario – Researchers at the University of Toronto, led by Dr. Philipp Maass, have developed "SIGNATURE," a new machine-learning method for mapping inter-chromosomal contacts between different chromosomes, also known as "kissing chromosomes," to explore the 3D structure of the human genome.

Published in Nature Communications, the study presents a supervised and unsupervised learning algorithm called Signature as a new approach to understanding chromosomal communication by identifying over 40,000 inter-chromosomal contacts across 53 human cell types.

Topological Anchors

The research shows that some chromosomal interactions overlap across different cell types and function as topological anchor communities. These anchors create a gradient of genome activity, together with other genomic features, such as gene expression.

The topological anchors are critical in shaping the genome's 3D structure, especially at the ends of chromosomes (telomeres). Acting as organizational hubs, they likely facilitate communication between genes and regulatory elements, which in turn affect cell-type-specific functions. Sixty-one topological anchors occur at the nuclear speckles - regions where many genes cluster. Dr. Philipp Maass, the principal investigator of this study, explains, “We observe many recurring interactions; it’s not random. We identified 61 topological anchors which reappear across numerous datasets and which seem to be important for genome structure across human cell types.

Rabl’s Configuration

This study provides new evidence for Rabl's configuration in human genomes, showing that it coexists with traditional chromosome territories. In the Rabl configuration, chromosome ends (telomeres) and central regions (centromeres) cluster on opposite sides of the nucleus. While this chromosomal structure was previously unclear in the human genome, Signature clearly visualized Rabl’s configuration, suggesting its role in genome stability and organization in human cell types.

Tissue-Specific and Sex-Specific Genome Interactions

Using Signature, the researchers identified that over 57% of interactions are cell-type-specific. Milad Mokhtaridoost,  first author of the study, notes, “Most interactions are unique to one cell type and bring genes into spatial proximity that are important for the cell’s function.”

The team also discovered differences and similarities in genome structure between male and female cells, including unique behaviour of the X chromosome in each sex. "X chromosomes behave differently in male and female cells," explains researcher Jordan Chalmers, emphasizing the sex-specific chromosomal interactions happening across the genome.

Implications for Disease Research

The compendium of genomic contacts provides a valuable resource to the scientific community and allows researchers to advance their understanding of chromosomal interactions within their own projects. Mapping these inter-chromosomal contacts in healthy genomes opens new doors for studying genomic re-organization in diseases like cancer. Dr. Maass adds, "…the next step is to dive into cancer genomes and other disease genomes to see what differs from the compendium of ‘healthy’ genomes that we currently have in our hands.”

Availability
The Signature code, documentation, and data are available on GitHub: https://github.com/MaassLab/Signature.

About the Research Team
This interdisciplinary research was funded by the Canadian Institutes of Health Research, NSERC, and the New Frontiers in Research Fund. It was conducted in collaboration with the University of Toronto, the Department of Molecular Genetics and the SickKids Research Institute, the Perelman School of Medicine, the University of Pennsylvania, Jena University Hospital, Friedrich Schiller University, Donnelly Centre, and the Department of Mechanical and Industrial Engineering, University of Toronto.

For more information, please contact the corresponding author: philipp.maass@sickkids.ca