AlphaGenome: Google’s DeepMind Cracks the Code of Life
Picture a future where researchers can access, in cookbook form, what makes up every human being. This is a dream… AlphaGenome will allow scientists to do this. The revolutionizing nature of such a tool will greatly enhance our understanding of DNA; help solve the crimes of genetic diseases; and expedite the search for new treatments for these diseases.
AlphaGenome is not just more artificial intelligence. This AI has been designed to investigate human DNA (the entirety of what makes you, you) at a very advanced level (until now). According to the research team, AI has the capability to examine 1 million coded letters of DNA at any given time, greatly assisting researchers in their investigation of an area of the human genome often termed the “dark genome.” Although 2% of total human DNA is responsible for providing instructions for your body to make proteins, 98% of DNA is, although not directly involved, responsible in various degrees for regulating how genes function. It is in the “dark” section of our genome where a lot of mutations causing health problems, such as, for example, high blood pressure, dementia, obesity and some types of cancer have been located.
Natasha Latysheva, a research engineer at DeepMind, describes “We see AlphaGenome as a tool for understanding what the functional elements in the genome do, which we hope will accelerate our fundamental understanding of the code of life”. Essentially, AlphaGenome is like a magnifying glass for looking at all of the DNA regions where we may have previously had no information about function.
Another reason why AlphaGenome is so unique is because it can predict the impact of even single letter changes in DNA. This is significant, as you can imagine that very small changes can have a huge impact on your health. AlphaGenome not only identifies where genes are located in the genome, but it also makes predictions regarding how the rest of the dark genome around the gene may be affecting it. AlphaGenome can tell, for instance, whether the gene is expressed or suppressed, and whether or not the body will splice together a variety of genes to produce different types of proteins.
Many researchers have already begun exploring these new options. The work of Dr. Gareth Hawkes at the University of Exeter illustrates this; he is using AlphaGenome to analyze genetic mutations that are linked to obesity and diabetes. “Most of the variants that we’re studying are considered part of the “dark genome.” He states, “They impact important biological processes we don’t fully understand yet. AlphaGenome helps predict what these variants are doing so we can test them in the lab”. Ultimately, these findings could lead to the discovery of novel therapeutic agents targeting these specific genetic alterations.
Another area where AlphaGenome may have significant applications is in the field of cancer research. Scientists have been able to use it to identify the mutations that drive the development of cancer as well as the mutations that might serve as targets for new treatment. Dr. Robert Goldstone, who works as a researcher within the Cancer & Genomics department at the Francis Crick Institute, referred to the AlphaGenome model as a major breakthrough in the development of his organization’s genomic AI program, and emphasized its capability to utilize the DNA sequence of an organism to predict how genes will be expressed.
Prof. Ben Lehner, a leading expert in synthetic biology, highlighted the AI’s potential for gene therapies and synthetic biology. By designing new sequences of DNA, scientists could create new ways to treat diseases at their root by incorporating synthetic biology. He adds, however, that while AlphaGenome is performing well in more than half a million experiments, it’s still “far from perfect.” Some predictions, especially those involving long-range gene interactions, need refining.
AlphaGenome’s approach is different from AI models like ChatGPT. While language models predict the next word in a sentence, AlphaGenome is a “sequence-to-function” model. It studies how changes in DNA affect its overall function, learning from publicly available experiments on human and mouse cells.
Following its launch for non-commercial purposes last year, AlphaGenome has been used by more than 3000 scientists to evaluate its potential. The AlphaFold project led to AlphaGenome being viewed as a step closer to leveraging artificial intelligence (AI) in the field of biology. Pushmeet Kohli, vice president of DeepMind, says, “I think we are at the start of a new era of scientific progress, and AI is going to enable a number of different breakthroughs”
Beyond advancing basic science, the uses of AlphaGenome have far-reaching implications for medical science, including the potential to accelerate the identification of rare genetic disorders, support the drug development process and contribute to generating personalised gene therapies. Additionally, it would enable synthetic biologists to generate completely new DNA sequences for therapeutic applications.
Despite the excitement, experts stress that there is still a long way to go. Accuracy in predicting gene regulation across different tissues and over long distances in DNA needs improvement. A neuron in the brain and a heart cell may have the same DNA letters, but how those instructions are read can be very different. Researchers hope to refine AlphaGenome further so it can handle these complexities.
Still, the reaction from the scientific community has been overwhelmingly positive. Many describe AlphaGenome as “an incredible feat” and “a major milestone.” Dr. Hawkes called it a “big leap” in understanding the dark genome, while Prof. Lehner emphasized how the combination of genomics, AI, and biomedical research could transform biology in ways never seen before.
AlphaGenome could be the start of a revolution in how we understand life at its most fundamental level. By reading the recipe for life in DNA, this AI from DeepMind may not only reveal the secrets behind genetic diseases and cancer but could also pave the way for future breakthroughs in synthetic biology and gene expression research.
The potential is enormous, but so is the challenge. Even as scientists celebrate the progress, they acknowledge that AlphaGenome is a tool, not a finished solution. With more research, refinement, and collaboration, it could become a cornerstone of 21st-century biology, unlocking the mysteries of the dark genome and giving humanity a new understanding of life itself.
