AlphaGenome Atlas maps billions of genetic changes with AI

The human genome contains approximately 3 billion DNA letters, creating more than 9 billion possible single-letter changes. Testing the effects of each change in a laboratory would be practically impossible. Google DeepMind's new AlphaGenome Atlas, available beginning today, gives scientists a comprehensive, searchable resource designed to accelerate understanding of the human genome.

Stowers Institute Bioinformatics Scientist and Zeitlinger Lab member, Melanie Weilert, served as a lead author on the AlphaGenome project. With her deep expertise in interpretating AI models, she helped build the AlphaGenome Atlas resource to ask one of biology's biggest questions: How does a cell know which genes to turn on and off?

"This is a very difficult problem because every cell type speaks a slightly different language, making it hard to know which rules are general," Zeitlinger said. "With AlphaGenome, we can quickly query many cell types and look for general patterns by which genes are activated and repressed."

Google DeepMind developed the technology behind the Atlas. Zeitlinger, who also leads the Stowers Institute's AI Initiative, helped connect its predictions to the biological processes that give cells their identities and allow them to function.

Pushmeet Kohli, Ph.D., VP Science, Google DeepMind and Chief Scientist, Google CloudAlphaGenome Atlas is a powerful example of how AI can expand human knowledge and advance scientific discovery. By making this resource widely available, we hope scientists around the world can use it to better understand the language of life and what happens when individual letters in the human genome change."

Scientists can access the resource through a web browser without writing code, allowing more researchers to explore genetic variation at a scale that was not previously possible.

"AlphaGenome Atlas is foundational research with the potential to have an impact across multiple areas of biology," said Avsec. "We worked with experts in the field, including Julia, whose biological insight helped us explore how the resource can map functional elements in the genome and reveal their roles at the molecular level."

How Stowers scientists helped reveal the regulatory "words" of the genome

Every cell in the human body contains essentially the same DNA, yet different cells use that information in very different ways. Short DNA sequences called motifs act as regulatory instructions, helping control which genes are active, when they are activated and how strongly they operate.

Zeitlinger and her team used AlphaGenome Atlas to analyze regulatory motifs across the genome and determine what they reveal about the proteins, called transcription factors, that control gene activity. The researchers categorized these regulatory signals by function, distinguishing transcription factors that change whether DNA is accessible from those that also activate or repress genes.

Conducting this type of analysis experimentally across thousands of sites and many different cell types would require enormous time and resources. By making genome-wide predictions available in one searchable resource, the Atlas allowed Zeitlinger's team to identify broader patterns in how genes are regulated and begin defining the general rules underlying the regulatory language of DNA.

"Having these motifs mapped at base-pair resolution across the genome and in many cell types gives us a searchable dictionary for non-coding DNA," Zeitlinger said. "By giving the scientific community access to these predictions, AlphaGenome Atlas can accelerate how we identify potentially disease-causing variants while helping us understand the fundamental rules by which genes are regulated."

"This collaboration demonstrates how Stowers scientists are helping shape emerging technologies, not simply adopting them," said Stowers Institute President and Chief Scientific Officer Alejandro Sánchez Alvarado, Ph.D. "By pairing deep biological knowledge with the capabilities of AI, researchers can ask questions at a scale that was not previously possible and create new opportunities to more clearly understand human health and disease."

The Atlas does not replace laboratory research. Instead, it can help scientists determine which variants and biological mechanisms should be investigated first, focusing experimental time and resources on the most promising questions.

"Tools such as AlphaGenome Atlas become most valuable when their predictions can be connected to meaningful biological questions," said Stowers Institute Scientific Director Kausik Si, Ph.D." Julia's work brings together deep expertise in gene regulation and computational biology to help move us from simply reading DNA sequence toward understanding the rules that control gene activity."

From billions of variants to focused biological questions

AlphaGenome Atlas contains thousands of molecular-effect predictions for each variant across hundreds of human cell types and tissues. These predictions contribute to the new AlphaGenome Variant Impact, or AVI, score. They also enabled researchers to identify and map recurring DNA motifs, short sequences where transcription factors bind to help control gene activity.

The AVI score brings together predictions from AlphaGenome, AlphaMissense and evolutionary conservation data. It gives researchers a single measure for ranking variants by their potential impact across protein-coding and non-coding regions of the genome. Researchers can then examine which molecular processes, including gene expression, RNA splicing and protein function, are predicted to be affected.

The collaborating institutions explored how the resource could support several areas of human genetic research. Scientists at the Broad Institute used the AVI score to prioritize a previously overlooked non-coding variant associated with an unsolved rare disease case. At the University of Exeter, researchers applied Atlas to genomic data from more than 54,000 UK Biobank participants, uncovering additional associations between rare noncoding variants and protein levels.

Source:

Stowers Institute for Medical Research