Google DeepMind Publishes AI-Powered Predictions For Effect of All 9 Billion Mutations to Human DNA
- Reference: 0185527512
- News link: https://tech.slashdot.org/story/26/09/08/1559238/google-deepmind-publishes-ai-powered-predictions-for-effect-of-all-9-billion-mutations-to-human-dna
- Source link:
> [3]AlphaGenome Atlas , as DeepMind calls the database, is a precomputed catalogue of what each substitution of a single DNA base is likely to do to the machinery that switches genes on and off. Until now researchers had to run such a model one variant at a time or had to test variants in the laboratory, a process that was painstakingly slow. It would have taken many human lifetimes to discover the consequences of all 9 billion possible single-letter mutations. The Atlas promises to make the job of biologists and medical researchers considerably easier, potentially speeding up the understanding of genetic diseases and the hunt for possible cures.
>
> Pushmeet Kohli, DeepMind's vice president for research and head of its AI for science team, told reporters on a briefing call that this was the first time any researcher in the world could reach a comprehensive map of human genetic variation "by simply opening a browser." Kohli also framed the release as helping to complete the unfinished business of the Human Genome Project, which in 2003 succeeded in mapping the entire human DNA sequence. "As the saying goes, we bought the book," he said, "but we did not understand how to read it."
>
> Atlas is available for non-commercial use from today through a website Google DeepMind has set up for it. The company said it would be available for commercial use through a licensing arrangement through Google Cloud "soon." Kohli said that Google DeepMind's sister company,ÂIsomorphic Labs, which is using AI for drug discovery, would have access to Atlas but that it would also require a commercial license for access. He did not specify exactly what the terms would be for commercial licensing. A paper describing the Atlas and how it was created is being released on bioRxiv, a repository for biomedical preprint academic papers.
[1] https://deepmind.google/blog/alphagenome-atlas-a-predictive-map-of-every-possible-dna-letter-change-in-the-human-genome/
[2] https://fortune.com/2026/09/08/google-deepmind-ai-predictions-9-billion-mutation-human-genome/
[3] http://alphagenome.google/atlas
Sadly 10 Billion of them (Score:3)
turn you into Aqua Man.
Re: (Score:1)
no mod points. coffee spilled. well done.
Best guess != labwork (Score:3)
> this was the first time any researcher in the world could reach a comprehensive map of human genetic variation "by simply opening a browser."
Not quite. This is the first time any researcher in the world can get an AI's best guess as to what each single substitution might do. All that "painfully slow" research of testing actual variants in the lab is, by contrast, real data.
Don't get me wrong, this is very, very cool, and will almost certainly lead to a lot of new discoveries - Along with a lot of false positives and dead-ends. I'm not even an AI hater, despite having serious reservations about alignment. This, however, is exactly the type of inquiry where AI hallucinates the worst: An open-ended "what if" question where the answer isn't allowed to be "I don't know".
Here's a quick sanity check every one of us should do occasionally - Go ask ChatGPT a subtle open-ended "why" question in your own field of expertise... And have a good laugh at how hilariously wrong the answer is.
Re: More false credit for the plagerism bots (Score:2)
Anything anybody does from this point on will have had some prediction by the AI which can be given some form of credit even if it wasn't even used. It could be completely wrong but simply highlight the gene or be in the general topic to get credit for contributing and proving AI is making discoveries.
I am surprised somebody hasn't made an AI psychic service yet. With enough people and targeted guessing it should catch many true believers! Remember that old scams sending out letters would narrow down the
Re: (Score:2)
> I am surprised somebody hasn't made an AI psychic service yet. With enough people and targeted guessing it should catch many true believers!
I am sure there are more, but I heard a company called 'openAI' has been selling this for quite some time already.
Re: (Score:2)
Thing is, even 51% accurate guess still can be useful.
Re: (Score:2)
> Anything anybody does from this point on will have had some prediction by the AI which can be given some form of credit even if it wasn't even used.
This is very true - in the same way that all steel refined after 1945 has artificial isotopes in it from the Trinity (and later) events. There is no recorking that bottle.
Best guess No Data (Score:2)
As you pointed out, I think a better question is how good is that DeepMind guess. If it moderately to high accurate, then it will greatly speed up planning of genetic research.
How did they verify those predictions? (Score:2)
Because if they did not, this could cause far more harm than good. Like cutting of entirely research directions because of faulty data in this database. That effect has happened often enough to be a major risk in biomed research.
Yeah ok, AI (Score:2)
Now explain why you think there are 2 e’s in the word seventeen.
So... (Score:2)
How did it tell if the "mutation" was a mutation or just a possible legit variant?
Re: (Score:3)
Patents. And AI can't be the inventor for patents, so by doing this, they're potentially dropping a huge land mine on patentability of genetics research. It's kind of awesome, at least from an "our genes are not for sale" perspective. :-D
How does ownership work? (Score:2)
Answer: Force.
> “Here's the thing. I don't know what you kids are up to, but I do know one thing. Laws are threats made by the dominant socioeconomic, ethnic group in a given nation. It's just a promise of violence that's enacted. And police are basically an occupying army, you know what I mean? You guys wanna make some bacon?” — Bud Cubby
Where's the funny? (Score:2)
Come on Slashdot, this story is a great opening for the Funny. But of course the most obvious and lowest hanging fruit is:
"AI predicts the AI won't hurt you."
Or perhaps:
"AI predicts you should invest all your money in AI. Do it NOW!"
How about:
"AI predicts that for $1/day you can feed a starving AI in China."
Notwithstanding [1]https://www.dwarkesh.com/p/ope... [dwarkesh.com] on the recent Hugging Face entertainment. (One of the AI cookbooks I read was from one of the primary architects of Hugging Face. But his prediction of "
[1] https://www.dwarkesh.com/p/openai-huggingface