With the help of AI, a potent antibiotic which can destroy some of the most dangerous drug-resistant bacteria in the world has been discovered.
This drug is named Halicin and works in a different way than current antibiotics. It’s the first of its type to be found by setting an AI searching digital libraries of pharmaceutical compounds.
According to tests, this antibiotic destroyed antibiotic-resistant strains, including two of three high-priority pathogens ranked as critical by the WHO.
Regina Barzilay, senior research on the project and a specialist in machine learning at MIT notes that this is a first in terms of antibiotic discovery.
James Collins, bioengineer on the team said that this is one of the most potent antibiotics discovered until now.
It has amazing activity against numerous pathogens that were resistant to current antibiotics.
What Is Antibiotic Resistance?
Antibiotic resistance happens when bacteria evolve and mutate to overcome the mechanisms of drugs that can kill them.
If there are no new antibiotics to fight off their resistance, 10 million lives worldwide can be at risk every year from bacterial infections by 2050, according to the O’Neill report.
In the discovery of new potent antibiotics, the MIT research trained a deep learning algorithm to find the sorts of molecules which can kill bacteria.
They achieved this by ‘feeding’ the program with info on the molecular and atomic features of almost 2500 drugs and natural compounds or how well or not the substance prevented the growth of E. coli.
When the algorithm learned the features of a good antibiotic, the scientists set it to work on a library of more than 6000 compounds for treatment of numerous illnesses.
The algorithm didn’t focus on potential antimicrobials, but on compounds which seemed effective but unlike the current antibiotics.
This elevated the chance for the drugs working in radical ways that a bug would have yet to develop resistance to.
What Did the Tests on Halicin Show?
According to the team, the tests on bacteria from patients showed that this antibiotic destroyed Mycobacterium tuberculosis which causes TB and strains of Enterobacteriaceae which is resistant to carbapenems, a group of antibiotics considered the last resort for these infections.
It also cleared out Acinetobacter baumannii and C difficile in lab mice.
Thanks to the digital database, the algorithm shortlisted 23 possible antibiotics, two of which appeared especially powerful.
Since they’re done with a computer, the experiments are faster and cost-effective.
Barzilay wants to use the algorithm to find antibiotics which are more selective in the bacteria they destroy or to design new antibiotics from nothing.