Researchers at Google DeepMind have unveiled their latest artificial intelligence tool and claim it will help scientists identify the genetic drivers of disease and ultimately lead to new treatments.
The alphagenome predicts how mutations interfere with how genes are regulated, when they are turned on, which cells in the body undergo changes, and whether their biological volume controls are set high or low.
Most common diseases that run in families, including heart disease and autoimmune disorders, as well as mental health problems, are linked to mutations that affect gene regulation, as are many cancers, but identifying which genetic glitches are responsible is not simple.
“We see AlphaGenome as a tool for understanding what functional elements in the genome do, which we hope will accelerate our fundamental understanding of the code of life,” DeepMind researcher Natasha Latysheva said at a press briefing on the work.
The human genome runs on up to 3 billion pairs of letters – the Gs, Ts, Cs and As that comprise the DNA code. About 2% of the genome tells cells how to make proteins, the basis of life. The rest orchestrate gene activity, containing key instructions that decide where, when, and how much individual genes are switched on.
The researchers trained AlphaGenome on public databases of human and mouse genetics, enabling it to learn the relationship between mutations in specific tissues and their effects on gene regulation. AI can analyze up to 1 million letters of DNA code at a time and predict how mutations will affect various biological processes.
The DeepMind team believes this tool will help scientists discover which strands of the genetic code are most essential for the development of particular tissues, such as nerve and liver cells, and pinpoint the mutations most important for cancer and other diseases. It could also underlie new gene therapies by allowing researchers to design entirely new DNA sequences – for example, switching on a certain gene in nerve cells, but not in muscle cells.
Carl de Boer, a researcher at the University of British Columbia in Canada who was not involved in the work, said: “AlphaGenome can identify whether mutations affect genome regulation, which genes are affected and how, and in what type of cell. A drug can then be developed to counteract this effect.
“Ultimately, our goal is to create models that are so good that we do not have to perform any experiments to confirm their predictions. While the alphagenome represents an important innovation, achieving this goal will require continued work from the scientific community.”
Some scientists have already started using alphagenomes. Mark Mansour, clinical professor of pediatric haemato-oncology at UCL, said this is a “step change” in his work to find genetic factors for cancer.
Gareth Hawkes, a statistical geneticist at the University of Exeter, said: “The non-coding genome is 98% of our 3 billion base pair genome. We understand the 2% quite well, but the fact that we’ve got an alpha genome that can predict what this other 2.94 billion base pair region is doing is a huge step forward for us.”
