How Genome Mapping Can Help Researchers Better Understand Disease
Google DeepMind's AlphaGenome and AlphaGenome Atlas tools assist researchers in decoding the human genome. By predicting the impact of genetic variants, these models help identify disease-causing mutations faster, accelerating the discovery of treatments for rare disorders and improving personalized medicine.
Tags
Quick summary
Google DeepMind's AlphaGenome and AlphaGenome Atlas tools assist researchers in decoding the human genome. By predicting the impact of genetic variants, these models help identify disease-causing mutations faster, accelerating the discovery of treatments for rare disorders and improving personalized medicine.
How Genome Mapping Can Help Researchers Better Understand Disease
The human genome serves as the body's instruction manual, containing approximately 3 billion base pairs of DNA that dictate how biological functions operate. While science has long been capable of reading this entire sequence, deciphering the specific semantics of the genetic code—particularly the non-coding "dark matter" that regulates gene expression and disease risk—remains a significant challenge. Determining which specific DNA changes cause diseases is often described as finding a needle in a haystack. To address this, Google DeepMind has developed AI tools, specifically AlphaGenome and the AlphaGenome Atlas, designed to accelerate the process of understanding disease mechanisms.
The Scale of Genetic Variation
The complexity of the genome lies in its vast potential for variation. The DNA sequence is composed of four bases, referred to as letters: adenine (A), thymine (T), cytosine (C), and guanine (G). These letters act as a biological alphabet. Because any of these letters can be replaced by one of the other three, there are 9 billion possible single-letter changes within the genome. Many genetic diseases are caused by just one of these specific letter changes.
However, conducting billions of physical studies to determine which changes make a difference is not feasible for scientists due to time and resource constraints. As noted by Google DeepMind research scientist Žiga Avsec, who oversees the genomics initiative, the sheer volume of potential variations requires a new approach to prediction.
AlphaGenome and the AlphaGenome Atlas
To tackle this challenge, Google DeepMind developed AlphaGenome, a model capable of taking a partial DNA sequence of 1 million base pairs in length and making thousands of predictions. By comparing these predictions for DNA changes, scientists can identify the most significant variants more quickly.
Building on this model, the team created the AlphaGenome Atlas, a massive dataset that allows scientists to decide what to research faster. When a patient with a rare disease is sequenced, the resulting data might reveal 4 million variants—4 million ways their DNA differs from the standard reference human genome. Even when narrowing the list to only include rare variants, researchers could still be left with hundreds or thousands of options.
The AlphaGenome Atlas addresses this bottleneck by providing a single, easy-to-read score: the AVI (AlphaGenome Variant Impact) score. This score instantly indicates the potential impact of a variant across the full genome, helping researchers prioritize and more easily find the most disruptive changes to follow up on.
From Reading DNA to Understanding Mechanisms
The utility of these tools extends beyond simple identification. According to Google DeepMind, tools like AlphaGenome move the scientific community from merely reading DNA to helping researchers better understand the molecular mechanisms behind a disease. This shift provides researchers with new leads to test, taking an important step toward the long-term goal of digitally simulating a complete human cell.
In the short term, the objective is to use AI to better understand the "grammar" of the genome and accurately predict the effects of mutations. In the long term, Google DeepMind aims to build on this foundation to create a fully functioning digital simulation of a human cell. This would allow researchers to computationally model disease mechanisms, understand rare genetic conditions, and test candidate therapies without the immediate need for physical experimentation.
Real-World Applications and Case Studies
Researchers are already utilizing these tools to advance science. Gareth Hawkes, a Medical Research Council fellow at the University of Exeter, used AlphaGenome Atlas to narrow down a list of 1 billion genetic variants. By focusing on the most impactful 1%, he identified 19 variant groups linked to body mass index to prioritize in his research.
Similarly, Laura Covill, a Broad Institute research fellow, used the AVI score to prioritize genetic variants in their research of unsolved rare diseases. AlphaGenome Atlas identified a critical variant in a gene linked to neurodevelopmental disorders, providing crucial supporting evidence that drove their subsequent research, including solving the case.
These examples illustrate how the tools can accelerate research that could help society improve the ability to diagnose and personalize treatment for patients suffering from rare or previously uncharacterized genetic disorders. Furthermore, the capabilities could speed up drug discovery pipelines by helping pharmaceutical researchers identify and validate high-probability therapeutic targets faster. Beyond healthcare, the model architecture could potentially help researchers tackle broader societal challenges, including breeding climate-resilient crops.
Limitations and Future Directions
Despite the advancements, there are limitations to the current technology. AlphaGenome was launched in June 2025, and the AlphaGenome Atlas was introduced in September 2026. While the model is state-of-the-art, it was trained mostly on data from whole tissues or cell lines. Google DeepMind acknowledges that individual cell types in those tissues have a very specific role and can malfunction in disease, which whole-tissue averages might obscure.
To address this, the team has launched AlphaGenome 2. This latest model tackles the key limitation of cell type specificity. It moves beyond tissue-level averages to predict how genetic variants disrupt these individual cell types, helping researchers pinpoint the cellular drivers of disease.
Additionally, Google DeepMind is actively researching better ways to measure how confident the model is in its predictions so researchers know how much to trust each result. From early analyses, variants with strongly predicted effects are quite good and achieve high levels of accuracy. However, the publisher emphasizes that anyone using AlphaGenome should interpret its predictions with other sources of information.
Conclusion
The integration of AI into genome mapping represents a significant shift in how researchers approach disease understanding. By leveraging AlphaGenome and the AlphaGenome Atlas, scientists can navigate the complexity of the 3 billion base pairs of human DNA more efficiently. These tools help transform the overwhelming task of finding a needle in a haystack into a manageable process of prioritization and prediction. As models like AlphaGenome 2 evolve to address cell type specificity and confidence measurement, the potential for digitally simulating human cells and personalizing treatments for rare disorders continues to grow.
Source: https://blog.google/innovation-and-ai/models-and-research/google-deepmind/alphagenome-disease-research



