IBM develops AI to invent new antibiotics – and it's made two already:
The IBM Research team created an AI system that's much faster at exploring the entire possibility space for molecular configurations. First, the researchers started with a model called a deep generative autoencoder, which essentially examines a range of peptide sequences, captures important information about their function and the molecules that make them up, and looks for similarities to other peptides.
Next, a system called Controlled Latent attribute Space Sampling (CLaSS) is applied. This system uses the data gathered and generates new peptide molecules with specific, desired properties. In this case, that's antimicrobial effectiveness.
But of course, the ability to kill bacteria isn't the only requirement for an antibiotic – it also needs to be safe for human use, and ideally work across a range of classes of bacteria. So the AI-generated molecules are then run through deep learning classifiers to weed out ineffective or toxic combinations.
Over the course of 48 days, the AI system identified, synthesized and experimented with 20 new antibiotic peptide candidates. Two of them in particular turned out to be particularly promising – they were highly potent against a range of bacteria from the two main classes (Gram-positive and Gram-negative), by punching holes in the bugs' outer membranes. In cell cultures and mouse tests, they also had low toxicity, and seemed very unlikely to lead to further drug resistance in E. coli.
Journal Reference:
Payel Das, Tom Sercu, Kahini Wadhawan, et al. Accelerated antimicrobial discovery via deep generative models and molecular dynamics simulations [open], Nature Biomedical Engineering (DOI: 10.1038/s41551-021-00689-x)
(Score: 0) by Anonymous Coward on Tuesday March 16 2021, @03:52PM (1 child)
Shithead AC poster here. I got nuthin - this actually seems like a good idea.
(Score: 1) by nostyle on Tuesday March 16 2021, @07:16PM
Oh sure. It's all rainbows and unicorns until you realize you are nearing the Singularity [wikipedia.org] wherein they will not allow you to live and not allow you to die.
I mean, when they say:
...be afraid. Be very afraid.
--
Magic 8 Ball said: "Outlook not so good". [Business did not listen.]
(Score: 2) by Runaway1956 on Tuesday March 16 2021, @04:09PM (2 children)
How much do teams such as this base their efforts on historical distributed computing?
Some of us have been doing distributed computing for a couple decades now. SETI@Home, Folding@Home, and all the BOINC projects. Millions of us around the world have been contributing computer time to dozens of projects.
It's great that IBM is making something newer, faster, and supposedly better. But, where was their starting point? How much did they rely on all that prior research into Alzheimers, cancers, AIDS, COVID, many communicable diseases, and so much more?
That doesn't even touch on all the computer science research that has advanced AI to current levels.
It would be cool to dig into a bibliography, and find that some project I contributed to was mentioned. Or, even a project that I didn't contribute to. That would still indicate that our efforts have been worthwhile.
“I have become friends with many school shooters” - Tampon Tim Walz
(Score: -1, Troll) by Anonymous Coward on Tuesday March 16 2021, @04:19PM (1 child)
Literally LOL at: "Some of us have been doing distributed computing for a couple decades now. SETI@Home"
That's like claiming you personally won the Cold War because you spent more of other people's money than the other guys did. The commies. Famous for spending other people's money AMIMOTHERFUCKINGRITE???
(Score: 0) by Anonymous Coward on Tuesday March 16 2021, @07:23PM
parent AC has probably never contributed anything to anyone anywhen for any reason
(Score: 0) by Anonymous Coward on Tuesday March 16 2021, @10:48PM
Let's call things what they are, not what marketdroids shot out of their collective arse.
(Score: 5, Interesting) by Azuma Hazuki on Wednesday March 17 2021, @01:11AM (2 children)
Now, yes, like a previous AC said, this isn't strong AI s much as a brute-force searcher with some memory to it. But if there is anything we humans ought to use technology for, it's to complement our weaknesses and do things we can't. In theory, we *could* do this with pencil and paper and large charts of chemical and pharmacokinetic data. In theory. In practice, the computer lets us do it literally millions of times faster.
Given antibiotic resistance is on the rise, this is literally a matter of life and death. I find it very interesting that the search is also focusing on small (peptide) molecules, as opposed to what the medical industry usually thinks of as antibiotics proper, which tend to be very large, especially in the case of macrolides, aminoglycosides, and the occasional outsider like vancomycin.
Novel mechanisms of action are also going to be important in the future. One of the newest antibiotics is called cefidericol, brand name Fetroja, and it's a fairly standard cephalosporin-class drug aside from having a catechol side chain that has massively strong affinity for iron. Since bacteria need iron (and, fun fact, a strong immune response causes low iron!), they will happily play tug-of-war with the cefidericol molecule's siderophore side-chain...yanking the drug inside themselves and letting it go to work doing its cephalosporin thing! Sneaky, sneaky medicine, and I predict we'll see new classes of siderophore medications soon.
I've said for a long time now that the solution to any given scientific problem lies one "tier" lower to the bare metal of reality, as it were: biological problems have chemical solutions, chemistry problems have physics solutions, etc. Antibiotic resistance is a biochemical problem, so it makes sense that the solution is a matter of the physics of chemistry :)
I am "that girl" your mother warned you about...
(Score: 0) by Anonymous Coward on Wednesday March 17 2021, @01:55AM (1 child)
Isn't macrolide ring smaller than anything longer than a tetrapeptide?
And aren't polypeptide antibiotics notorious for their toxicity?
https://en.wikipedia.org/wiki/Polypeptide_antibiotic [wikipedia.org]
(Score: 2) by Azuma Hazuki on Wednesday March 17 2021, @03:29AM
Yes, polypeptides are notorious for toxicity. I've never yet seen colistin ordered and I hope I never do.
Regarding macrolides, here's the structure of erythrymycin: https://en.wikipedia.org/wiki/Macrolide#/media/File:Erythromycin_A.svg [wikipedia.org] Not the hugest molecule ever, but the size of this vs. a tetrapeptide is going to depend on what the 4 peptides are. By comparison, here's vancomycin, which is holy-crap huge: https://en.wikipedia.org/wiki/Vancomycin#/media/File:Vancomycin.svg [wikipedia.org]
I haven't thoroughly digested the paper, but I get the impression they're finding compounds like teixobactin rather than new members of canonical classes like beta-lactam, aminoglycoside, etc., with this search.
I am "that girl" your mother warned you about...