A new computational tool maps genome change, helping researchers see DNA in 3D

New research from the Wisconsin Institute for Discovery is tackling a complex packing problem. Thanks to the development of a powerful new computational tool introduced in a study published earlier this year in the journal Genome Research, scientists can better investigate how genomes fit into the tiny confines of a cell nucleus, how they are repackaged across different biological dimensions and how that influences gene expression and disease risk.

Genomes can be massive and often need to be repackaged as cells develop and specialize, across different developmental stages, disease states and at other varying points in time. Gene expression must be tightly choreographed since changes in 3D genome structure have been linked to shifts in gene activities that involve diseases like cancer and genetic disorders.

Intrigued by this puzzle, researchers Sushmita Roy and Da-Inn Lee, who graduated from the University of Wisconsin–Madison last May, developed a computational method called Tree-Guided Integrated Factorization (TGIF) that creates a mathematical model of the DNA folding, based on a machine learning technique called matrix factorization. Roy and Lee hope the tool will help systematically examine how changes to DNA across different dimensions influence traits that range from controlling hair color to genetic diseases

“One of the fundamental questions in mammalian genomics is how DNA is packaged inside the nucleus, so that the relevant parts are available for the cell to read while the other parts are stowed away depending upon the context; this is especially difficult since much of it is non-coding,” says Sushmita Roy, Professor of Biostatistics and Medical Informatics and WID faculty. 

Non-coding DNA hasn’t historically been considered very useful because it doesn't have the code for crucial proteins that do things like build or repair tissue, fight infections and control chemical reactions. However, scientists are learning that non-coding DNA has important functions, like instructing or regulating which genes need to be active and therefore what proteins should be made.

“What we still don’t fully understand is which pieces of DNA control which genes. That’s why studying how DNA is packaged in the nucleus could be key to unlocking new insights into gene regulation in normal cellular function and disease risk,” says Roy.

Previous methods looked for how these changes occur across pairs of time points or conditions. The Roy Lab wanted to build a tool that could track changes across multiple time points and account for complex structures.

When testing the analytical framework, they found a strong correlation in how large and small changes in 3D genome structure relate to shifts in gene expression, especially in genes involved in timepoint-specific functions. They also looked at the boundaries between tightly interacting genome regions and found that stable boundaries were often linked to genetic variants associated with disease.

“This is a crucial step toward understanding the genotype to phenotype relationship,” says Roy, “which plays a key role in how organisms function under different environmental conditions. For example, we found that conserved boundaries were associated with single nucleotide polymorphisms implicated in cardiovascular disease.”

With this new tool, the WID team is adding a powerful piece to the puzzle of gene regulation — one that could help scientists better understand not just how the genome works, but how its shape shapes us.