AI Predicts 3D Structures of Nearly Every Protein Known to Science

Science

August 02, 2022

An AI Has Predicted the 3D Structure of Nearly Every Protein Known to Science

Hi, friends!

Welcome to the future of medicine. And botany. And biotechnology. And... anything else science-related, really.

Just last week, a company called DeepMind announced that it coded an artificial intelligence to predict the 3D structure of nearly every protein known to science. “The entire protein universe,” it's calling it. This is an absolutely massive step forward for pretty much any sort of scientific research because the interplay between proteins and molecules dictates basically everything biological about the world.

In 1957, it took 22 years of research to decode the first 3D protein structure. It was a slow, steady process since then. Last year, that number rose up to 350,000. Now, we’re at more than 200 million. Essentially all of them. More details in our top story linked above.

Right below, you can take a lovely journey through the reason why a 10-year-old Martian landing changed science writer Eric Mack’s life forever. It’s a wonderful, thought-provoking read.

If you scroll down even farther, you will find my massive mixed bag of featured science stories this week. Have any of you read the book The Three-Body Problem? I have not, but I, regrettably, read the entire Wikipedia page on it and therefore know the story. (Yes, I’m one of those people. But, in my defense, the Wikipedia article was so impressive that I will probably still read the book.) Anyway, scientists located a three-star system that is reminiscent of the plot of this book, for anyone who read it the normal (much better) way.

We’ve also got the scoop on NASA’s Mars rock return mission, ESA’s plan to plunge a spacecraft into Venus’ hellish atmosphere and so much more!

📧The Mailbag📧

John asks: If we could switch positions in space, putting Hubble where the JWST is and vice versa, how much better [for Hubble] and worse [for JWST] might be the images we see from the respective scopes?

Hi John! This is a really interesting question -- and the short answer here is that if the JWST was launched to Hubble’s spot in Earth’s orbit, and Hubble was sent to the JWST’s spot at L2 long ago, both would probably never have worked at all.

Here’s the long answer.

Starting with JWST, this telescope is a big deal because it uses special instruments to detect what’s called infrared light. Infrared light isn’t part of the visible spectrum, yet light emanating from deep space phenomena appears to us on Earth -- and in Earth’s general vicinity -- as infrared. But the thing about infrared light is it’s better thought of not as “light” but as a heat signature. So, as the JWST is trying to capture all those incredibly precise infrared light heat signatures from deep space, it has to be totally shielded from other types of heat, aka the sun. Otherwise, those unwanted heat signals would act like noise in the data and interfere with all of the JWST’s equipment. Thus, the JWST’s positioning at L2 is absolutely crucial to its mechanism -- scientists carefully selected this spot because here, the telescope is always on the side of the Earth not facing the sun and safe from such radiation.

With regard to Hubble, this telescope doesn’t specialize in infrared signatures. It can detect some, but it’s more focused on visible wavelengths, so it doesn’t really need L2’s special heat protection aspect. When scientists sent Hubble up, their main concern was getting a telescope above our planet’s atmosphere for super-clear pics. It probably could’ve gone there when it launched decades ago, but that’d have been unnecessarily complex and expensive to accomplish.

However, the reason I say Hubble wouldn’t have worked had we sent it to L2 is a bit of a funny (read: stressful) story. When Hubble was first sent up, it simply didn’t work. The pics it sent back were blurry. Eventually, NASA realized the issue was with the ‘scope’s lenses, so the agency bravely sent astronauts to Hubble in LEO to fix it in midair. And they did. The rest is history. At L2, repairs probably wouldn’t have been possible. And Hubble might never have really worked.

Yes, that means we probably can’t manually repair the JWST if needed.

Apologies for the lack of brevity on this one -- but don't forget, whenever you like, send science questions, thoughts, comments, chats, space-based worries and philosophical ponderings to my email or message me on Twitter! And if you're enjoying this column, please do forward it to your friends!

Enim scientia et astra!

-M