Google DeepMind AI Model Reads DNA’s Recipe for Life

A powerful AI model developed by Google’s DeepMind is being hailed as a potential turning point in genetics.

Researchers reckon it could dramatically accelerate understanding of how DNA influences disease, cancer, and drug discovery.

The model, called AlphaGenome, is designed to read and interpret the human genome – the complete set of genetic instructions that governs how the body grows, functions, and responds to illness. Scientists say it offers new insight into why tiny variations in DNA can raise the risk of conditions such as high blood pressure, obesity, dementia, and diabetes.

While its creators stress that the technology is still evolving, independent experts have described it as “an incredible feat” and “a major milestone” for genomic science.

Unlocking the ‘dark genome’

The human genome consists of around three billion letters of genetic code, represented by A, C, G, and T. Only about 2% of this code directly forms genes that instruct the body to make proteins. The remaining 98% has long been labelled the ‘dark genome’ because its role is harder to interpret.

Despite its name, this dark genome is far from irrelevant. It plays a crucial role in controlling when, where and how genes are switched on or off, and it contains many of the mutations linked to disease. Understanding how this regulatory DNA works has been one of the biggest challenges in modern biology.

AlphaGenome is designed to tackle this problem by analysing up to one million letters of DNA at a time. It can identify where genes are located and predict how surrounding non-coding regions influence gene expression and gene splicing, the process that allows a single gene to produce different proteins.

Crucially, the model can also predict the biological impact of changing just a single letter in the DNA code, something that has traditionally required lengthy laboratory experiments.

Implications for disease and drug discovery

One of the most promising applications of AlphaGenome is in identifying which genetic mutations actually cause disease, rather than merely being associated with it. This distinction is critical for developing effective treatments.

In the longer term, AlphaGenome could also support advances in synthetic biology, including the design of new DNA sequences for gene therapies. Such applications could allow scientists to correct faulty genetic instructions rather than simply treating symptoms.

The model was described in detail in the journal Nature and was made available for non-commercial use last year. Since then, around 3,000 scientists worldwide have begun using it in their research.

Obesity, diabetes, and cancer research

Cancer is an area where the technology could have significant impact. AlphaGenome has already been used to distinguish between mutations that actively drive cancer growth and those that are biologically incidental. This could help researchers focus on the mutations most likely to respond to targeted treatments.

Dr Robert Goldstone, head of genomics at the Francis Crick Institute, describes the model as a “major milestone in the field of genomic AI” and highlights its “ability to predict gene expression from DNA sequence alone.”

Not perfect, but transformative

Despite the enthusiasm, researchers caution that AlphaGenome is not a finished solution. Prof Ben Lehner, head of generative and synthetic genomics at the Wellcome Sanger Institute, told the BBC that his team tested the model in more than half a million experiments and found it performed very well, but he emphasises its limitations.

“It’s far from perfect,” he says, noting that the model struggles with predicting long-range gene regulation, where DNA regions influence genes located more than 100,000 letters away. Improving accuracy across different tissue types is another challenge, as the same DNA sequence can behave very differently in brain cells compared with heart cells.

Still, Lehner says the broader implications are profound. “It’s a really exciting time with three areas where the UK is world-leading – genomics, biomedical research, and AI – combining to transform biology and medicine.”

A new era for AI in biology

AlphaGenome follows DeepMind’s earlier breakthrough with AlphaFold, an AI system that predicts the three-dimensional structure of proteins. That work earned the DeepMind team the Nobel Prize for Chemistry in 2024 and has already reshaped biological research.

Pushmeet Kohli, vice president of science and strategic initiatives at Google DeepMind, believes AlphaGenome represents another step towards AI-driven scientific discovery. “I think we are at the start of a new era of scientific progress, and AI is going to enable a number of different breakthroughs,” he says.

While refinements are still needed, many researchers believe AlphaGenome has the potential to fundamentally change how scientists read, interpret and ultimately rewrite the genetic code that underpins human life.

Last month, Google revealed plans for its first-ever automated research lab, backed by a £5 billion investment that could boost the UK’s economic trajectory.