Google DeepMind released a new AI system called AlphaGenome to help researchers find the genetic causes of disease. The tool predicts how mutations change gene activity in different body tissues. It can analyze up to 1 million letters of DNA code at once.
How AlphaGenome Works
The human genome contains 3 billion pairs of letters. Only 2% of these letters tell cells how to make proteins. The remaining 98% controls when and where genes turn on or off. AlphaGenome focuses on this larger portion.
According to The Guardian, DeepMind trained the system on public databases of human and mouse genetics. The AI learned connections between mutations in specific tissues and their effects on gene regulation.
Most common diseases that run in families have been linked to mutations affecting gene control. These include heart disease, autoimmune disorders and mental health problems. Many cancers also involve such mutations. But finding which genetic changes cause disease remains difficult.
Potential Medical Applications
Natasha Latysheva, a DeepMind researcher, said the team sees AlphaGenome as a tool for understanding functional elements in the genome. The system could help scientists map which genetic sequences are essential for developing particular tissues like nerve and liver cells.
The tool might also support new gene therapies. Researchers could design DNA sequences to switch on certain genes in specific cell types but not others. Carl de Boer at the University of British Columbia noted that the system identifies whether mutations affect genome regulation and in what cell types. Drugs could then be developed to counteract these effects.
Scientific Reception
Some researchers have already started using AlphaGenome. Marc Mansour, a clinical professor at UCL, called it a step change in his work finding genetic drivers for cancer.
Gareth Hawkes at the University of Exeter said the non-coding genome makes up 98% of the 3 billion base pair genome. He added that having AlphaGenome make predictions about this region represents a big step forward. De Boer noted that achieving fully reliable predictions will require continued work from the scientific community.
