The DNA sequence contains more than just the instructions needed to produce proteins. The order of its four letters — A, T, C, and G — also determines how the molecule bends, how much it can deform, and how it interacts with other cellular components. These physical properties influence fundamental biological processes and can shape how the genome itself is regulated.

An international study led by the Institute for Research in Biomedicine (IRB Barcelona) now presents the most complete characterization to date of this hidden “physical code” of DNA. Published in Nature Communications, the work analyzed 2,080 unique hexamers using molecular dynamics simulations. To pull it off, 14 research groups from the Ascona B-DNA Consortium spent three years simulating two replicas of 190 double-stranded DNA fragments, each 20 base pairs long — generating roughly 250 terabytes of atomic-resolution data in the process.

Why Shape Matters as Much as Sequence

“To function, DNA must bend, open, and interact with numerous proteins. Therefore, knowing how the nature of the sequence conditions its physical behaviour is essential for understanding genome regulation and evolution, beyond the information contained in genes,” said Dr. Modesto Orozco, who led the research.

Dr. Federica Battistini, a co-author on the study, said the sheer scale of the simulations let the team capture behavior that’s usually out of reach. “We have been able to observe both the typical behaviour of each sequence and exceptional movements that normally fall outside the reach of simulations,” she said. The atlas shows that DNA fragments with similar composition can behave very differently depending on the order of their bases, and the long duration of the simulations allowed researchers to catch rare events, including transient openings and partial unwindings of the double helix.

A Possible Footprint in Evolution

The team also investigated whether these physical properties may have shaped genome evolution over time. Studying regions of the genome that don’t code for proteins, researchers found that the most frequent sequences tend to better resist small changes — even when one letter is substituted, their shape and movements are barely altered. Though the trend is moderate, the results suggest that evolution may have favored sequences capable of preserving the physical properties necessary for DNA to function.

The data generated have been made available to the scientific community following the FAIR principles, and the researchers say the resource could be used to build simplified computational models and AI systems capable of representing DNA and chromatin at larger scales. The consortium included institutions from Spain, France, the United Kingdom, the United States, Germany, Poland, Switzerland, Uruguay, and Lithuania.

The study was published in Nature Communications (DOI: 10.1038/s41467-026-74390-5), with funding from the European Commission through BioExcel and the MDDB project, Spain’s Ministry of Science, Innovation and Universities, the Carlos III Health Institute, the European Regional Development Fund, and the Generalitat de Catalunya through AGAUR.