Scientists use ML to find an antibiotic able to kill superbugs in mice
1–10 of 93 posts
Re: Scientists use ML to find an antibiotic able to kill superbugs in mice
#2This paper shows what can be done when you carefully run an ML program alongside a wet lab experimental program tailored to feed back into the ML program. The results end up far more interesting than some recent "ML aided drug discovery" papers in that they actually discovered a drug that functions very differently than known antibiotics (AB).
Even though the structures that came out look AB-like, they work differently than known ABs, probably by disrupting the pH gradient across the cell membrane. Other ABs might do the same as part of their activity, but work better under different conditions than this one. The result is an innovative structure, and a molecule that can hit resistant strains.
Combining ML and wet lab is the real way we'll get new drugs. You need to regularly check in with a high content ground truth or you'll come up with either uninteresting or useless results. I'm a bit biased though, that's what we do at my company ;)
Re: Scientists use ML to find an antibiotic able to kill superbugs in mice
#3> That is an especially pressing challenge in the development of new antibiotics, because a lack of economic incentives has caused pharmaceutical companies to pull back from the search for badly needed treatments. Each year in the U.S., drug-resistant bacteria and fungi cause more than 2.8 million infections and 35,000 deaths, with more than a third of fatalities attributable to C. diff, according to the the Centers for Disease Control and Prevention.
How big does the market have to be to commercially viable for research and development? Nearly 3M potential patients at a couple hundred dollars per course is nearing a $1B/year.
Re: Scientists use ML to find an antibiotic able to kill superbugs in mice
#4> That is an especially pressing challenge in the development of new antibiotics, because a lack of economic incentives has caused pharmaceutical companies to pull back from the search for badly needed treatments. Each year in the U.S., drug-resistant bacteria and fungi cause more than 2.8 million infections and 35,000 deaths, with more than a third of fatalities attributable to C. diff, according to the the Centers…
The second-order costs avoided by treatments developed so innovatingly could be included in a "value to society" estimation.
"Acknowledgements" lists the grant funders for this federally-funded open access study.
"A Deep Learning Approach to Antibiotic Discovery" (2020) https://doi.org/10.1016/j.cell.2020.01.021
> Mutant generation
> Chemprop code is available at: https://github.com/swansonk14/chemprop
> Message Passing Neural Networks for Molecule Property Prediction
> A web-based version of the antibiotic prediction model described herein is available at: http://chemprop.csail.mit.edu/
> This website can be used to predict molecular properties using a Message Passing Neural Network (MPNN). In order to make predictions, an MPNN first needs to be trained on a dataset containing molecules along with known property values for each molecule. Once the MPNN is trained, it can be used to predict those same properties on any new molecules.
Re: Scientists use ML to find an antibiotic able to kill superbugs in mice
#5This method might be generalizable to prions.
Re: Scientists use ML to find an antibiotic able to kill superbugs in mice
#6This paper shows what can be done when you carefully run an ML program alongside a wet lab experimental program tailored to feed back into the ML program. The results end up far more interesting than some recent "ML aided drug discovery" papers in that they actually discovered a drug that functions very differently than known antibiotics (AB). Even though the structures that came out look AB-like, they work different…
Can you help me understand what is the difference between what you are doing and what people like Certara and Simulations Plus are doing? The simulation software market is really fascinating
Re: Scientists use ML to find an antibiotic able to kill superbugs in mice
#7> That is an especially pressing challenge in the development of new antibiotics, because a lack of economic incentives has caused pharmaceutical companies to pull back from the search for badly needed treatments. Each year in the U.S., drug-resistant bacteria and fungi cause more than 2.8 million infections and 35,000 deaths, with more than a third of fatalities attributable to C. diff, according to the the Centers…
At least one disincentive is that if you do find an amazing new antibiotic effective against certain strains of antibiotic resistant bacteria, antibiotic stewardship means the medical community will try and use it only where necessary to slow any adaptation to the new drug. That makes your potential patient population much smaller.
Re: Scientists use ML to find an antibiotic able to kill superbugs in mice
#8Thank goodness it was "aided by machine learning". Otherwise, would anybody be reading this article ;)?
Re: Scientists use ML to find an antibiotic able to kill superbugs in mice
#9...in mice.
It would be nice if we spent as much time, money and attention on figuring out prevention. Inadequate hygiene infrastructure (like toilets) in developing areas is part of the problem here.
But addressing that isn't as exciting to people as finding a cure for a super bug. If we really want to fix this, that needs to change.
Re: Scientists use ML to find an antibiotic able to kill superbugs in mice
#10This paper shows what can be done when you carefully run an ML program alongside a wet lab experimental program tailored to feed back into the ML program. The results end up far more interesting than some recent "ML aided drug discovery" papers in that they actually discovered a drug that functions very differently than known antibiotics (AB). Even though the structures that came out look AB-like, they work different…
Not just in drug discovery but in most interesting industries. Using ML as either a human/cyborg aid or ML+real world ground truth is a secret superpower that I'm surprised more people don't know about. I'm glad they don't.