Google DeepMind Unveils AlphaGenome AI to Decode Genetic Mutations

The tool is currently available for non-commercial, research-focused use via an API. While it is not approved for clinical diagnosis, DeepMind states that AlphaGenome can help researchers identify mutations that may be linked to disease.

DeepMind has launched AlphaGenome, a new artificial intelligence model designed to predict how individual mutations in human DNA affect their functions.

The model focuses particularly on non-coding regions of DNA, areas that do not produce proteins but play key roles in regulating gene activity.

The company says AlphaGenome represents a step forward in genome research, offering predictions on how genetic variants, both common and rare, affect biological processes. 

These insights may help researchers unravel the complexities of diseases linked to gene regulation, such as cancer.

Understanding mutations in these regions has historically been difficult due to their complexity and the limited interpretive power of existing tools. AlphaGenome attempts to address this by analysing DNA sequences up to one million base pairs in length and producing detailed predictions across multiple cell types and conditions.

According to DeepMind, the model was trained on public datasets from large-scale genomics consortia including ENCODE, GTEx, 4D Nucleome, and FANTOM5. These datasets contain information on human and mouse cell types, enabling AlphaGenome to interpret a wide range of genomic activity.

The AI model builds upon DeepMind’s earlier work on Enformer and complements AlphaMissense, which focuses on mutations within the 2% of the genome that encodes proteins.

In contrast, AlphaGenome is tuned to interpret the remaining 98% of the genome, the non-coding segments, where many disease-associated mutations are thought to reside.

AlphaGenome consolidates various analytical functions into a single model, reducing the need for multiple tools traditionally used to assess the impact of gene mutations. In one early use case, researchers used the model to predict how a known leukemia-related mutation could activate a disease-driving gene, a finding that aligned with prior laboratory studies.

The tool is currently available for non-commercial, research-focused use via an API. While it is not approved for clinical diagnosis, DeepMind states that AlphaGenome can help researchers identify mutations that may be linked to disease.

The company indicated that future updates may expand the model’s capabilities to additional species and biological systems.

Stay tuned for more such updates on Digital Health News