ARC Innovation at Sheba Medical Center and the Icahn School of Medicine at Mount Sinai announced on Monday a partnership with NVIDIA to crack the majority of the human genome that remains poorly understood.
The collaboration aims to apply large language model technology to decode more than 98 percent of DNA that does not code for proteins, potentially unlocking new pathways for disease prevention, diagnosis, and treatment, according to Israeli financial news outlet Globes.
The initiative involves an investment of tens of millions of dollars, with each of the three partners assigning five to seven people to work full-time on the project.
NVIDIA will contribute computing power, infrastructure, algorithms, and AI teams, while Sheba Medical Center will provide clinical data, and Mount Sinai will supply genetic information from 11,000 genomes as part of its ongoing Million Health Discoveries Program that aims to sequence one million patient genomes over five years.
Why 98 Percent of Human DNA Remains Poorly Understood
While the first human genome was sequenced in 2000, scientists quickly discovered that protein-coding genes represent only about 1 to 2 percent of human DNA.
The remaining 98 percent, once dismissed as junk DNA, is now understood to include crucial regulatory elements that control gene expression and play major roles in diseases including cancer, heart disease, and autism.
Prof. Gidi Rechavi, the scientific leader of the project on behalf of Sheba, explained that more than 98 percent of genes serve as part of a complex orchestra of genes whose role is probably to very precisely and delicately control the genes that express proteins.
The complexity of non-coding DNA stems from its role in three-dimensional genome organization, enabling long-range interactions that regulate cellular function across vast genomic distances.
These regulatory elements include gene promoters, transcription factor binding sites, enhancers, transposable elements, and topologically associating domain boundaries that collectively orchestrate when and where genes activate throughout human development and daily physiological processes.
Understanding this hidden regulatory code represents one of the greatest challenges in modern genomic medicine.
Did you know?
Scientists once dismissed 98 percent of human DNA as junk with no biological function, but research now reveals these non coding regions contain crucial regulatory elements that control gene expression and play major roles in diseases including cancer, heart disease, and autism.
How Will AI Language Models Decode Genomic Information
The partnership plans to apply large language model technology, similar to systems used for processing human language, to identify patterns and relationships within the vast expanse of non-coding DNA sequences.
These AI systems excel at finding subtle patterns in massive datasets, making them potentially valuable for detecting regulatory relationships that traditional genomic analysis methods might overlook.
Prof. Eyal Zimlichman, Director of ARC at Sheba, stated that only by combining the unique strengths of the three partnering organizations can they solve one of the toughest challenges in healthcare that touches at the very core of how the human body works.
The project will progress incrementally, first deciphering the activity of regulatory genes in healthy people, then examining common diseases, and eventually expanding across all areas of health and disease.
NVIDIA's expertise in developing AI infrastructure and algorithms that process enormous computational workloads positions the company to accelerate genomic discovery at scales previously impossible for medical research institutions working independently.
The hospitals will define the project as successful if they can learn something new about disease pathways within approximately two years of launch.
What Resources Are Partners Contributing to the Project
NVIDIA will provide the computing power, infrastructure, algorithms, and dedicated AI teams necessary to process and analyze the enormous datasets generated from genomic sequencing.
The company's graphics processing units and AI accelerators have become the industry standard for training large language models, and this expertise will now apply to biological sequences rather than text.
Sheba Medical Center in Tel Hashomer will contribute clinical data collected from its patient population, with the project physically conducted at the hospital's facilities in Israel.
Mount Sinai will supply genetic information from 11,000 genomes already sequenced as part of the Million Health Discoveries Program, which the Icahn School of Medicine launched with the Regeneron Genetics Center in 2022.
This ambitious initiative aims to enroll one million Mount Sinai patients over five years, making it one of the largest genetic sequencing efforts globally and providing researchers with unprecedented ethnic and racial diversity in genomic data.
Each organization will assign five to seven full-time staff members to the collaborative effort, ensuring dedicated expertise from medical, genetic, and computational disciplines.
ALSO READ | What Makes This New Breast Cancer Drug Different from Others
Which Diseases Could Benefit From Genome Breakthrough
Non-coding DNA variations have established roles in numerous human diseases based on existing research, with growing evidence from cancers, Parkinson's disease, autism spectrum disorders, Rett syndrome, neurofibromatosis, Friedreich's ataxia, amyotrophic lateral sclerosis, and frontotemporal dementia.
Mutations in enhancer elements of specific genes are linked to disease progression, while alterations in regulatory regions contribute to cancer development and neurodevelopmental disorders.
Prof. Gidi Rechavi, who heads the Cancer Research Center at Sheba Medical Center, noted that 98 percent of the genome previously ignored is now recognized as containing critical regulatory and functional elements.
Decoding the variability in these non-coding regions represents the key to next-generation diagnosis and therapy, according to researchers involved in the project.
The ability to identify disease-causing variations in regulatory elements could enable precision medicine approaches that target the root causes of gene dysregulation rather than merely treating symptoms.
Understanding how non-coding variants contribute to disease susceptibility, progression, and treatment response could revolutionize how clinicians approach prevention strategies and therapeutic interventions across virtually every medical specialty.
Who Owns the Intellectual Property From This Research
All intellectual property created from the collaboration will remain in the hands of Sheba Medical Center and Mount Sinai, while the AI model itself will be placed in the public domain according to Prof. Yitshak Kreiss, director general of Sheba Medical Center.
However, the two hospitals will have initial and unique access to the AI model before broader public availability, giving them competitive advantages in applying genomic insights to patient care and research.
This arrangement balances the goals of advancing scientific knowledge through open access while protecting the hospitals' investment and providing a return on their substantial commitments.
The decision to make the underlying AI model publicly available signals recognition that accelerating genomic medicine requires collaboration across the global research community rather than proprietary silos.
By retaining intellectual property rights to specific discoveries while sharing the foundational technology, Sheba and Mount Sinai position themselves to monetize clinical applications and therapies developed from the research while enabling other institutions to build upon the work.
This hybrid approach to intellectual property management may serve as a model for future large-scale collaborations between healthcare institutions and technology companies in the rapidly evolving precision medicine landscape.
Comments (0)
Please sign in to leave a comment