Imagine , You are a scientist , A new antibiotic needs to be found to fight terrible diseases . How will you find it ？
Usually , You have to test a lot of different molecules in the laboratory , Until we find a molecule with the necessary bactericidal properties . You may find competitors who are good at killing bacteria , Just realize that you can't use them , Because they also prove toxic to humans . It's a very long time , Very expensive process , And it can be very troublesome .
however , If the opposite is true , You can just type in the properties you want to find on your computer , Then let the computer design the ideal molecule for you ？
This is a IBM The general method used by researchers , That is, using molecular designs that can be automatically generated for new antibiotics AI System . Published in 《 Natural Biomedical Engineering 》 Last one In the new paper , The researchers detailed how they used it to quickly design two new antimicrobial peptides （ Small molecules that kill bacteria ）, They are effective against a variety of different pathogens in mice .
Usually , This molecular discovery process will take scientists years .AI The system did it in a few days .
That's good news , Because we urgently need faster ways to produce new antibiotics .
Why antibiotic resistance is a huge problem
When new antibiotics are introduced , They can have huge and even life-saving results . since 1928 Since the discovery of penicillin in , It ushered in a modern era of antibiotics , We're starting to rely on them to treat killers like tuberculosis , And in the process of cesarean section or joint replacement to ensure our safety .
But experts Warning , We are now entering the post antibiotic era , At that time, the antibiotics we had became almost useless . We're going through crops , The overuse of antibiotics in the treatment of farm animals and humans caused this crisis .
The more we overuse antibiotics , The more chance bacteria have to adapt to our drugs , And evolved into antibiotic resistant superbacteria , So that our drugs don't work .
According to the Pew Charitable Trust （Pew Charitable Trusts） Of The latest report ,Covid-19 The pandemic exacerbated the problem . Doctors are even more inclined to prescribe antibiotics unnecessarily . Even though Covid-19 It's a viral disease , also Antibiotics don't work on the virus , But doctors have been providing these drugs to patients , To prevent them from secondary infection when they are in hospital - Even when they don't know if the patient is infected .
Now , As you take the time to read this article , One person in the United States will be Infection and Death , This infection is due to our overuse of antibiotics and can no longer be effectively treated . In this year , There will be global 70 Ten thousand people died of drug-resistant infections . A major report of the United Nations Warning , To 2050 year , The annual death toll could increase to 1000 ten thousand , Unless we make some fundamental changes .
The reason why big pharmaceutical and biotech companies don't develop new antibiotics , Because it takes many years and a lot of money to develop . Most new compounds fail . Even if it works , And the benefits are small ： Antibiotics don't sell as much as the drugs you take every day . For many pharmaceuticals The company is , Financial incentives don't exist yet .
however , If you can use AI Get the job done quickly and cheaply ？
IBM Of AI How the system works
IBM The new AI The system depends on something called a generative model . For the simplest understanding of it , We can break it down into three basic steps .
First , The researchers started with a huge database of known peptide molecules .
then ,AI Extract information from a database and analyze patterns , To find out the relationship between molecules and their properties . You might find , When a molecule has some structure or composition , It tends to perform some function . This makes it possible to “ Study ” The basic rules of molecular design .
Last , Researchers can tell... Exactly AI What they want a new molecule to have . They can also enter restrictions （ for example ： Please keep the poison low ！）.AI Use this information about ideal and bad features , And then we design new molecules that meet the parameters . Researchers can choose the best of them , And started testing mice in the lab .
just as IBM One of the co authors of the paper AleksandraMojsilović Tell me that ,“ You just turn the knob , You get molecules that satisfy the properties .”
IBM Researchers claim , Their method is better than other leading methods in designing new antimicrobial peptides 10％. They found that they were able to design two new antimicrobial peptides that were highly effective against a variety of pathogens , Including Klebsiella pneumoniae, which is resistant to a variety of drugs （K.pneumoniae）, This bacterium is known to cause infections in hospital patients . It is gratifying that , When tested in mice , These peptides are less toxic , It's an important signal about its safety （ Although not all of them are right for mice, they can eventually be extended to humans ）.
In an article In a blog post ,IBM The researchers point out that , Even though they are interested in AI How the system can potentially accelerate antibiotic discovery and excite antibiotic resistant bacteria , But they also want the system to be more widely used . They imagined , It can help scientists “ Finding and designing better drug candidates , For more effective disease drugs and therapies , Materials that absorb and capture carbon to help fight climate change , Materials for smarter energy production and storage , wait .”
It's not about this AI The system can solve any of these problems by itself . But it proposes a computing strategy to solve the problem , Can bring really exciting benefits , And it could save a lot of lives .
Click on the title