Google DeepMind has launched a new artificial intelligence (AI) tool called AlphaGenome Atlas, a free database that includes information on 9 billion possible changes to the human genetic code. This tool estimates how each change might affect different tissues and cellular processes, offering scientists a quick way to assess the broader impacts of these genetic variations. AlphaGenome Atlas builds on a previous DeepMind tool called AlphaGenome, which predicts how changes in DNA sequences affect proteins, cells, and human health. The new version aims to make this technology more accessible by providing a user-friendly web portal, reducing the need for specialized bioinformatics knowledge. The database is massive, containing 1 petabyte of data, and is available for free, helping to reduce the computational demands on researchers. Experts have acknowledged the potential of AlphaGenome Atlas. Greg Findlay, a group leader at the Francis Crick Institute in London, described it as a valuable resource. However, Tuuli Lappalainen, a professor in genomics at KTH Royal Institute of Technology and the New York Genome Center, noted that while it's a step forward, the tool still cannot answer all major genetic questions. She emphasized that the technology is not entirely new but has improved in precision compared to earlier models. AlphaGenome Atlas enables researchers to explore how a single change in a DNA sequence might affect up to a million surrounding genetic elements, capturing the complex regulatory networks of genes. A preprint study by DeepMind showed that the tool's AlphaGenome Variant Impact score can effectively distinguish between disease-causing mutations and harmless genetic variations in clinical data. Despite these advancements, a preprint study led by Katie Pollard, director of the Gladstone Institute of Data Science and Biotechnology, highlighted that AlphaGenome sometimes underestimates the impact of certain mutations, especially when the changes are in regulatory regions far from the genes they influence. Lappalainen advised researchers to use these AI predictions with caution and emphasized the need for "wet-lab" experiments—traditional laboratory research—to confirm AI findings. While such experiments are time-consuming, they remain essential for validating AI predictions, as the field of genomics still faces data limitations that require further experimental confirmation.