Google DeepMind has unveiled an artificial intelligence (AI) invention, AlphaGenome, which its developers claim could have a “transformative impact” on medicine discovery by predicting how DNA mutations behave. The technology also holds promise for identifying genes linked to specific conditions and uncovering the causes of rare diseases.
Since its launch last June, thousands of researchers globally have utilised AlphaGenome to aid studies into “neurodegenerative diseases, infectious diseases and cancer.” The model functions by predicting how variants or mutations within DNA influence a range of biological processes that regulate genes.
This capability could significantly assist researchers in pinpointing the precise causes of diseases, enhancing genetic testing, and accelerating the development of new treatments. Furthermore, it is expected to deepen scientists’ understanding of the human genome – the complete set of DNA instructions found in every cell.
AlphaGenome was trained using both human and mouse genomes. Writing in the journal Nature, its developers revealed the program can simultaneously predict 5,930 human or 1,128 mouse genetic signals. These predictions either matched or surpassed the performance of existing state-of-the-art models in 25 out of 26 evaluations.
Natasha Latysheva, a research engineer at Google DeepMind, highlighted numerous applications where AlphaGenome “could have a transformative impact,” including drug discovery. She explained: “The idea here is by combining large genetic association studies, such as those from UK Biobank, with AlphaGenome predictions, scientists could better pinpoint the genes and the cell types associated with the particular trait or disease. This could add another piece of the puzzle for the discovery of drug targets and ultimately, the development of new drugs.”
Ms Latysheva further elaborated on its potential in cancer research: “In cancer, patients can harbour many different mutations simultaneously, and it’s often challenging to differentiate between the large numbers of passenger non-causal mutations from the causal driver mutations. A model like AlphaGenome could help prioritise down lists of variants to those most likely to actually be functional and causal and contributing to the illness.”
She added that AlphaGenome could also help identify the potential causes of rare diseases and has “interesting applications” in gene therapy. “The idea here is that if you have a powerful DNA sequence to function model, you can actually start to use that model to design entirely new DNA sequences with specific desired properties. For example, you could try to design a sequence that activates certain gene only in nerve cells, but not in muscle cells.”
AlphaGenome has been available to researchers via an application programming interface (API) since June 2025, allowing external software systems to interact with it. Pushmeet Kohli, vice president of science and strategic initiatives at Google DeepMind, stated that approximately 3,000 scientists from 160 countries have since made one million API calls.
The AlphaGenome model and its rates are now being released for non-commercial research, with a commercial version currently undergoing early testing. Mr Kohli noted that researchers from prominent academic institutions, including UCL, are already utilising the model “to advance research into areas including neurodegenerative diseases, infectious diseases and cancer.”
Reflecting on the broader scientific landscape, Mr Kohli drew an analogy, stating that while “proteins are only one chapter of the biological story, “if proteins are the ingredients of life, then DNA is the recipe.”
He added: “While the Human Genome Project gave us the Book of Life, reading it remained a challenge. We have the text, but we are still deciphering the semantics. Understanding the grammar of this genome, what is encoded in our DNA and how it governs life, is the next critical frontier for research.”
Professor Ben Lehner, head of generative and synthetic genomics at the Wellcome Sanger Institute in Cambridge, praised AlphaGenome as “a great example of how AI is accelerating biological discovery and the development of therapeutics.” He emphasised: “Identifying the precise differences in our genomes that make us more or less likely to develop thousands of diseases is a key step towards developing better therapeutics. AlphaGenome and models like it that help decipher the regulatory code of our genome will make it much easier to do this.”
Professor Lehner confirmed that the Wellcome Sanger Institute has tested AlphaGenome with 500,000 new experiments, finding it performs “very well.” However, he cautioned that “AI models are only as good as the data used to train them. Most existing data in biology is not very suitable for AI – the datasets are too small and not well standardised.” He concluded that “the most important challenge right now is how to generate the data to train the next generation of even more powerful AI models. We need to do this fast, cost effectively and in a way that both the data and the resulting models are available for everyone to use.”
Dr Robert Goldstone, head of genomics at the Francis Crick Institute, hailed AlphaGenome as “a major milestone in the field of genomic AI.” He explained: “This level of resolution, particularly for non-coding DNA, is a breakthrough that moves the technology from theoretical interest to practical utility, allowing scientists to programmatically study and simulate the genetic roots of complex disease.” Dr Goldstone concluded that while “AlphaGenome is not a magic bullet for all biological questions, it is a foundational, high-quality tool that turns the static code of the genome into a decipherable language for discovery.”
