of Welcome to the Sanger Institute, Google Deep Mindand Google.org They formed a five-year artificial intelligence consortium to create genomic datasets for machine learning models used in biological research.
The consortium, announced at the AI x BIO conference, will focus on generating large biological datasets in areas where existing data is limited or not yet structured for advanced AI systems.
The research will combine the genomics research and data generation capabilities of the Wellcome Sanger Institute with the AI expertise of Google DeepMind. Google.org is supporting the effort, but the amount of funding was not disclosed in the announcement.
The founding organizations plan to establish a broader consortium and invite additional collaborators. No additional participating organizations have been announced at this stage.
The dataset is intended to support AI models that can be trained and tested on genomic and molecular data. The consortium’s five-year program will focus on generating data suitable for machine learning, rather than adapting existing datasets after they are created.
Focus on data generation for 5 years
The Wellcome Sanger Institute and Google DeepMind are already working together through research collaborations, a joint AI fellowship in genomics, and projects focused on building AI research capacity, including in low- and middle-income countries.
The new consortium will move that relationship into a larger data-generating program. The partners say this research will support AI models used to predict biological processes and analyze areas of life science research that are not yet well covered by existing datasets.
Researchers at the Wellcome Sanger Institute are already using AI across genomics, including research that integrates information from genes, RNA, and proteins. AI is also used by Sanger researchers to analyze large datasets and support experimental design and interpretation.
Dr. Julia Wilson, Chief Innovation and Impact Officer at Wellcome Sanger Institute, said: “Combining Sanger’s expertise in producing world-leading datasets with Google DeepMind’s leadership in artificial intelligence gives us the opportunity to accelerate the generation of biological data specifically designed to power the development of fundamental AI models. Through this consortium, we aim to create a resource that will be widely shared with the community to enable groundbreaking scientific discoveries and have broad impact across the life sciences.”
The role of Google DeepMind in biological AI
Google DeepMind’s role in the consortium focuses on AI for science and model development. This announcement positions the partnership around a dataset designed from the ground up for AI training and evaluation.
“Together with the Sanger Institute, we aim to build the data backbone needed to decipher the complexity of biological processes. Ultimately, this could accelerate scientific discovery and open entirely new frontiers for researchers around the world,” said Dr. Pushmeet Kohli, vice president of AI for Science at Google DeepMind.
Anna Koivuniemi, Head of Google DeepMind Impact Accelerator, added: “Tackling biology’s most important challenges requires collaboration across disciplines, disciplines, and institutions. We are excited to partner with the Sanger Institute to power AI in genomics and ensure this data helps accelerate breakthroughs that benefit humanity.”
Open access data infrastructure
Google.org’s involvement is focused on supporting an open-access data infrastructure for future biological AI models. The announcement does not provide details about how the dataset will be accessed, managed, or published.
said Leslie Yeh, Director of Google.org Scientific Progress. “Over the past decade, we have seen how deep learning can transform our understanding of complex biological challenges. With this new consortium, Google.org is supporting the open-access data infrastructure needed to power the next generation of biological AI models. By accelerating the integration of large genomic datasets, we aim to help the global research community achieve life-saving scientific breakthroughs.”
The next step is to form a broader consortium and launch a five-year dataset program. The founding partners have not yet named any additional collaborators or set a publication schedule for the dataset.
