Finding New Antibiotics with Machine Learning



Don't you break down for me? What the machine learning approach that they used here. And what kind of advanced does this represent took this machine learning model that they made and they traded on about twenty five hundred molecules and use that to train binary classification models to predict probability of whether it new compound would inhibit the growth of e coli or not and then turn to the truck library Library of six thousand compounds. That are ready in human clinical development for wide variety of indications and at this point the compared several different models and after narrowing down there's molecules and actually predicting toxicity using different neural networks. They've came up with this particular molecule and Howson and then thirdly lastly in the process they went on to apply machine learning motto after iteration and optimization too much broader set zinc fifteen data set with over a billion a half structures and under machine learning side. What's key here's the deep learning network that the US didn't really rely on any information about the chemical structures of molecules. It actually really built new representations called for years. A lot of people represented molecules with these fingerprint factors reflected things like presence or absence of functional. Groups are descriptors and comparable properties but relying on known fingerprints. Didn't really work that well. And that's why you know. A lot of the old antibiotic screening process gives you a lot of the same classes of molecules over and over again and what they did here they actually have these fingerprint descriptors that were built from. Scratch well you know. What you're describing is still a fingerprint. Right into dimensional vector to describe molecules. I think perhaps what's different is at deep learning approach. You can try to infer what the right descriptors should be. That's the hallmark of all of the deep. Learning approaches for drug design is at the end deep learning in general that recall even when we're just talking about conventional neural nets for image recognition the ideas that CNN's for image. Recognition versus classical computational vision. Is that in the classical approach? That person's defines what the right features are and so similarly. You know it's interesting that you can feed any representation of molecule into computer which parts of the interesting ones. You can have just like old school computer version. You could have a human being say all these in the important ones but a a beauty of DNA approach which is Ucla but also in many precursor works. That helps understand. What are the key aspects? And what are the interesting ones? And that is really. I think the big difference between what you can get in modern deep learning with machine learning versus classical machine learning with like random forest or something like that right so deep learning helps us figure out what we don't know versus focusing only on what we already know or what we think we know what makes Alison an attractive candidate for further research and development. What are some of the properties that they discovered without a doubt certainly a really potent inhibitor of e? Coli and you know. Further investigation showed that Halston has strong growth. Inhibitory effects on a wide following spectrum of pathogens. They tried it on. C. Money and which is one of the highest priority pathogens that is urgently required for in terms of antibiotics and then more. Interestingly it was even able to eradicate equal I persist yourselves that remained after episode and treatment so pretty strong efficacy and pretty low talked based on their screen. It also checks the box of something that is really structurally divergent from conventional antibiotics. And so certainly a very powerful new class of antibiotics that could potentially be strong candidate for further development. Yeah the fact that they found the antibiotic they showed it worked in vitro they showed. It worked in Vivo. And then they also did some experiments together this mechanism of action suggesting that Howson selectively disrupts the Ph potential across the bacterial membrane this saps the Proton motive force which is like the battery of the cell so all antibiotics. It's disrupting an essential cellular function but this appears to be a distinct and new function. That's being targeted super elegant work such a complete well rounded story. Yeah well I mean I think one of the things that really stands out here. Is that full stack of experiments. That they've done where it goes. All the way from looking at they might see in a dish to going through mice and one of the appealing things about studying antibiotics. And this often even pertain. Santa Viral that the animal models are pretty good with something like alzheimers. On the far extreme where animals are generally not very good. And so it's appealing that one could go. Do all of this. You know. Probably not requiring huge budget and therefore get something on the other side of that. Looks kind of intriguing beyond initial discovery? Can this kind of machine? Learning based approach be applied to other aspects of either earlier late stage drug development. Yeah it's it's brought Tommy with the fun thing about it. There's a reason why it's a broad topic because there's a broad range of things you can do. I mean you could talk about identifying targets and there's a lot of work to do there an extra shooting novel targets. It's really interesting. Time to go after novel targets. You could talk about identifying leads. Basically what's been done here that identification leads and then the testing of them. These compounds are leads but presumably. They're not drug like so they have to be optimized so there's these methods helping lead optimization and then along the way hopefully you'd want to also be screening for talks and so there's a ton of methods that are getting really surprisingly accurate basically the beautiful thing about a machine learning approach like this is at the approach for the most part is pretty agnostic to what you're predicting and that the processes you're building up can be useful one last thing and this is maybe the holy grail dream is that if you're predicting a lot of properties for a lot of different systems with a whole bunch of molecules in some multitask like framework where one model is predicting all of it. You can learn from all of it and that you develop even though you might not have a lot of data in any single project or any single area here. The sum of all this data now is huge and helps to regularise your predictions to make them less over fit and more

Coming up next