Designer DNA could change how diseases are treated

Google DeepMind's AlphaGenome can read long DNA sequences, including non-coding regions, and predict how subtle mutations affect biological processes; the article reports it could be used to design synthetic DNA to switch genes on or off in specific tissues.

Google DeepMind has released AlphaGenome, an AI tool that interprets long strands of DNA and predicts how subtle mutations influence biological processes. The programme reads sequences up to a million base pairs and includes non-coding regions that make up most of the genome. Researchers say it can show how changes drive disease and help identify mutations linked to health effects. The article notes that using synthetic DNA to control gene activity in specific tissues is not currently possible with existing gene therapy. Key details: - AlphaGenome can interpret DNA sequences up to a million base pairs and predict the effects of subtle mutations and variants. - The tool can analyse non-coding or so-called "junk" DNA, which the article reports accounts for about 98% of the genome and helps regulate genes. - DeepMind researchers say AlphaGenome could generate short stretches of synthetic non-coding DNA to act as tissue-specific switches in a gene therapy delivery vector. - The programme may help pinpoint which mutations cause disease and identify beneficial variants that could inform future treatments. - DeepMind previously released tools to interpret gene effects, and AlphaGenome extends that work into non-coding regions. Summary: The article presents AlphaGenome as a research tool that expands the ability to interpret the genome beyond protein-coding regions and could enable design of synthetic DNA for tissue-specific gene control. Practical application in clinical settings and wider use remain undetermined at this time.