{"id":18629,"date":"2024-05-23T20:18:58","date_gmt":"2024-05-23T17:18:58","guid":{"rendered":"https:\/\/www.liberalglobe.com\/?p=18629"},"modified":"2024-05-23T20:18:58","modified_gmt":"2024-05-23T17:18:58","slug":"alphafold-googles-artificial-intelligence-revolutionizes-drug-discovery","status":"publish","type":"post","link":"https:\/\/www.liberalglobe.com\/?p=18629","title":{"rendered":"AlphaFold: Google’s artificial intelligence ‘revolutionizes’ drug discovery"},"content":{"rendered":"\n

It all started in 2020, when the AlphaFold system blew scientists away with its ability to predict how proteins work. It was the science news of the year, according to Science magazine.<\/p>\n\n\n\n

Now, the artificial intelligence of DeepMind, a London-based company now owned by Google, is going a step further: the new version can now predict the interactions of proteins with other molecules, such as DNA and RNA, a capability that is expected to be widely used in design of new drugs.<\/p>\n\n\n\n

And that’s because the AI system does in a few minutes what until now biologists needed months or years to achieve.<\/p>\n\n\n\n

The new version presented in Nature (“Accurate structure prediction of biomolecular interactions with AlphaFold\u20093<\/a>“) “is simply revolutionary,” Frank Ullman, a biochemist at the Francis Crick Institute in London who had the opportunity to test Alpha Fold3, commented on the journal’s website.<\/p>\n\n\n\n

“I’m sure every structural biology and protein biochemistry research group will immediately adopt the system,” Julien Bergeron, a biologist at King’s College London, told Science magazine (“Powerful new AI software maps virtually any protein interaction in minutes<\/a>“).<\/p>\n\n\n\n

Structure and function<\/strong><\/p>\n\n\n\n

Nothing works properly in cells without proteins, long chains of amino acids that fold into complex, three-dimensional shapes. The shape of each protein determines how it acts, whether it is a structural protein or an enzymatic protein that regulates metabolism.<\/p>\n\n\n\n

That is why the ability to predict their structure is expected to play an important role in biology, biochemistry and ultimately the development of drugs and treatments.<\/p>\n\n\n\n

Until recently, however, their three-dimensional structure could only be calculated with demanding and extremely time-consuming methods such as X-ray crystallography and cryo-electron microscopy, methods that do not always provide answers.<\/p>\n\n\n\n

By 2020 when the first version of AlphaFold was presented, scientists had managed to determine the structure of only 170,000 proteins. DeepMind’s system took just a few months to predict the structures of the roughly 200 million proteins known to science, from humans to bacteria.<\/p>\n\n\n\n

Diffusion<\/strong><\/p>\n\n\n\n

In its original form, AlphaFold could only predict the “static” structures of free proteins.<\/p>\n\n\n\n

Now it can predict how their structure changes when they interact with other molecules, such as DNA, RNA or other protein molecules.<\/p>\n\n\n\n

Ullman, for example, a biochemist at the Francis Crick Institute, has already used AlphaFold3 to study proteins involved in DNA replication when cells divide.<\/p>\n\n\n\n

“With these new capabilities, we can design molecules that bind to a specific region of a protein and predict how strong the binding will be,” Demis Hassabis, the British-Cypriot founder of DeepMind, told a news conference attended by Reuters.<\/p>\n\n\n\n

“It’s a critical step if you want to design drugs,” he said.<\/p>\n\n\n\n

The new version incorporates a multitude of changes and improvements, the company’s researchers say in their publication.<\/p>\n\n\n\n

Among other things, it uses the “diffusion” approach that many synthetic image tools such as DALL-E are based on, an approach that improves results by adding and removing statistical noise.<\/p>\n\n\n\n

Restrictions on access<\/strong><\/p>\n\n\n\n

Unlike previous versions of AlphaFold, which anyone could download and use, AlphaFold3 will only be available through DeepMind, specifically through the company’s launched “AlphaFold server.”<\/p>\n\n\n\n

Researchers will be able to upload protein sequences and lists of biomolecules that those proteins interact with. Results appear within minutes.<\/p>\n\n\n\n

However, the server does not provide the structures of proteins that interact with drugs, and each researcher will only be able to upload 10 protein sequences per day.<\/p>\n\n\n\n

DeepMind admits that these restrictions were imposed for the sake of Isomorphic Labs, a DeepMind spin-off that develops new drug treatments.<\/p>\n","protected":false},"excerpt":{"rendered":"

It all started in 2020, when the AlphaFold system blew scientists away with its ability to predict how proteins work. It was the science…<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1168,2960],"tags":[1624,5468,864,5467,2742,5466,815,3549],"class_list":["post-18629","post","type-post","status-publish","format-standard","hentry","category-science-technology","category-technology","tag-ai","tag-alpha-fold3","tag-artificial-intelligence","tag-deepmind","tag-drug","tag-drug-treatments","tag-google","tag-treatment"],"_links":{"self":[{"href":"https:\/\/www.liberalglobe.com\/index.php?rest_route=\/wp\/v2\/posts\/18629","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.liberalglobe.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.liberalglobe.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.liberalglobe.com\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.liberalglobe.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=18629"}],"version-history":[{"count":1,"href":"https:\/\/www.liberalglobe.com\/index.php?rest_route=\/wp\/v2\/posts\/18629\/revisions"}],"predecessor-version":[{"id":18631,"href":"https:\/\/www.liberalglobe.com\/index.php?rest_route=\/wp\/v2\/posts\/18629\/revisions\/18631"}],"wp:attachment":[{"href":"https:\/\/www.liberalglobe.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=18629"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.liberalglobe.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=18629"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.liberalglobe.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=18629"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}