January 28, 2026

expert reaction to paper on Google DeepMind’s AlphaGenome

A study published in Nature looks at Google Deepmind’s AI AlphaGenome tool for predicting function from DNA sequences. 

Professor Kristian Helin, CEO of The Institute of Cancer Research, London, said:

“AlphaGenome represents a major advance in computational genomics, matching or exceeding the performance of current state-of-the-art models for predicting functional elements directly from DNA sequence.

“The model shows strong predictive accuracy across key regulatory features, including RNA splice-site usage, the effects of genetic variants on gene expression, and chromatin accessibility. While it performs well in predicting overall gene expression levels, capturing cell type–specific regulation remains an important challenge.

“Despite these limitations, AlphaGenome establishes a powerful foundation for sequence-to-function modelling. In a rapidly evolving field, this work marks a significant step forward and helps lay the groundwork for more comprehensive approaches to functional genome interpretation.

“Even at this early stage, AlphaGenome has the potential to change how researchers generate hypotheses, prioritise experiments, and design more targeted and informative studies. Its immediate impact may be incremental, but its longer-term significance for the field is substantial.

“As seen previously with AlphaFold, the full impact of such approaches is likely to unfold over time. With continued methodological advances and the integration of increasingly rich experimental datasets, AlphaGenome could ultimately transform how we interpret genomes, understand disease mechanisms, and translate genetic variation into biological and clinical insight.”

Dr Xianghua LI, Lecturer in Medical and Molecular Genetics, King’s College London, said:

“The press release gives an accurate summary of the research. The main achievement is that this new AI can make many genetic predictions simultaneously, whereas most current tools handle only one. Bringing these abilities together in one system is a helpful advance.

“When we look at each prediction, this AI performs as well as the best existing tools, but not better. For important medical tasks, current AI models are still not reliable enough for patient care. Some ‘best performing’ tools often overstate the risks of certain genetic changes. These predictions are not yet ready for use in clinics.

“From a scientific point of view, the model does not uncover new findings about genes or biology. Some tough challenges remain, such as predicting very rare genetic variants. We still do not have enough good data for these cases.

“Still, this work is valuable as a starting point and a solid base for future progress. I hope the next version of Alphageome will show us deeper insights into how one genetic change, or several together, can affect many traits at once or in sequence. This understanding is crucial for learning about biology and making accurate predictions about genetic variants.”

Dr Robert Goldstone, Head of Genomics at the Francis Crick Institute, said:

“DeepMind’s AlphaGenome represents a major milestone in the field of genomic AI. This level of resolution, particularly for non-coding DNA, is a breakthrough that moves the technology from theoretical interest to practical utility, allowing scientists to programmatically study and simulate the genetic roots of complex disease.

“The model performs exceptionally well on tasks that might be expected to be governed by rigid ‘grammatical’ rules written in the DNA, such as splice site prediction. In these areas, it is poised to immediately replace older standard tools. And one of the most remarkable demonstrations is its ability to predict gene expression from DNA sequence alone. Whilst not perfect, given that gene expression is influenced by complex environmental factors that the model cannot see, achieving the level of accuracy demonstrated, based solely on ‘local’ DNA rules, is an incredible technical feat.

“AlphaGenome is not a magic bullet for all biological questions, but it is a foundational, high-quality tool that turns the static code of the genome into a decipherable language for discovery.”

Professor Ben Lehner, Head of Generative and Synthetic Genomics, Wellcome Sanger Institute, Cambridge, said:

“AlphaGenome is a great example of how AI is accelerating biological discovery and the development of therapeutics. Identifying the precise differences in our genomes that make us more or less likely to develop thousands of diseases is a key step towards developing better therapeutics. AlphaGenome and models like it that help decipher the regulatory code of our genome will make it much easier to do this.

“As we have come to expect from Google Deepmind, AlphaGenome is a great piece of engineering that brings together ideas developed by many different scientists into a model that sets the standards. At the Wellcome Sanger Institute we have tested AlphaGenome using over half a million new experiments and it does indeed perform very well. 

“However, AlphaGenome is far from perfect and there is still a lot of work to do. AI models are only as good as the data used to train them. Most existing data in biology is not very suitable for AI – the datasets are too small and not well standardized. The most important challenge right now is how to generate the data to train the next generation of even more powerful AI models. We need to do this fast, cost effectively and in a way that both the data and the resulting models are available for everyone to use. 

“Interestingly, AlphaGenome was entirely developed in the UK by a very international team. This isn’t luck but rather a direct consequence of years of government, charity and industry investment in science and attracting talent from across the world. It’s a really exciting time with three areas where the UK is world-leading – genomics, biomedical research and AI – combining to transform biology and medicine. This magic combination of large-scale data generation and AI needs to be very strongly supported in the UK and Europe to make sure that the breakthroughs continue to happen here and also that they are used to benefit everyone.”

Dr Fergal Martin, Eukaryotic Annotation Team Leader at EMBL’s European Bioinformatics Institute (EMBL-EBI), said:

“AlphaGenome shows how far AI has come in helping us understand our genomes. It demonstrates that we can now predict many important biological features directly from DNA, without needing to run new experiments for every genome. This makes it possible to interpret the differences between human genomes faster. Further down the line, models like AlphaGenome could extend beyond humans to help interpret the DNA of plants, animals, and microbes that haven’t been studied in detail before.

“AlphaGenome is one example among many AI models being developed in this area. The speed of progress will depend on sharing these models openly so the wider scientific community can test, improve, and build on them.…”

Professor Aldo Faisal, co-director of the School of Convergence Science in Human and Artificial Intelligence at Imperial College London, said:

“First of all, it is great to see progress in genome AI is being made by using public research data spanning national borders, I hope that the model will remain in the open and public domain going forward. The model sits in broad ecosystem of genome AI and is trained on major public functional genomics resources and uses an architecture that’s a sensible evolution of what we already know works. The team reports broad benchmark coverage (e.g., outperforming specialized models on many tasks). As the impact of this model will be on research and development workflows, I would treat the results as promising rather than final until 1. more groups reproduce the findings and 2. we see robust, pre-specified evaluation panels.”

Professor Rivka Isaacson, Professor of Molecular Biophysics in the Department of Chemistry at King’s College London, said:

“This work is an exciting step forward in illuminating the ‘dark genome’. We still have a long way to go in understanding the lengthy sequences of our DNA that don’t directly encode the protein machinery whose constant whirring keeps us healthy. There are so many interwoven possibilities, and complex feedback mechanisms, that I doubt the whole thing will ever be fully untangled. AlphaGenome gives scientists whole new and vast datasets to sift and scavenge for clues.”

‘Advancing regulatory variant effect prediction with AlphaGenome’ by Demis Hassabis & Pushmeet Kohli et al. was published in Nature 16:00 UK time on Wednesday 28 January 2026.

DOI: 10.1038/s41586-025-10014-0

Declared interests

Dr Robert Gladstone: No declarations of interest. There is a DeepMind lab at the Francis Crick Institute, but Robert has not worked on AlphaGenome.

Dr Fergal Martin: Fergal is involved in collaborative research with Google DeepMind. Fergal is not an author on, and was not involved in the development of the AlphaGenome model. EMBL-EBI has ongoing research collaborations with Google DeepMind, including work related to AlphaFold.

Prof Ben Lehner: has some research funding from Google DeepMind / a small collaboration with them. Ben is not an author on, and was not involved in the development of, the AlphaGenome model.

Professor Rivka Isaacson: No COIs.

Prof Aldo Faisal: I have no financial conflicts of interest with respect Google or Deepmind (unless pension funds type links).

For all other experts, no reply to our request for DOIs was received.