AI and machine learning have discovered a powerful antibiotic that can kill some of the most dangerous drug-resistant bacteria in the world. The drug is the first of its kind to be found by AI analyzing vast digital libraries of pharmaceutical compounds.
The tests carried out by the Massachusetts Institute of Technology (MIT) showed that the drug has wiped out a range of bacterial antibiotic-resistant strains, including Acinetobacter baumannii and Enterobacteriaceae, two of the three high-priority pathogens ranked as “critical” by the World Health Organization. Researchers say that it is one of the more powerful antibiotics that has been discovered to date.
The researchers first trained a “deep learning” algorithm to identify the kinds of molecules that kill bacteria in order to find new antibiotics. To do so, they fed the program information on the atomic and molecular characteristics of nearly 2,500 drugs and natural compounds, and how well or not the substance blocked the bug E coli’s growth.
Having learned what molecular characteristics made good antibiotics, the scientists set it to work on a library of over 6,000 compounds under investigation to treat various human diseases. Instead of looking for any potential antimicrobials, the algorithm focused on compounds that looked effective but unlike antibiotics already in existence. This boosted the chances of the drugs working in radically new ways to which bugs had yet to develop resistance. It took the algorithm a matter of hours to evaluate the compounds, and develop some promising antibiotics.
Tests on patient-collected bacteria showed that halicin killed Mycobacterium tuberculosis, the bug that causes TB, and carbapenem-resistant strains of Enterobacteriaceae, a group of antibiotics considered to be the last resort for such infections. Halicin also cleared infections of mice with C difficile and multidrug-resistant Acinetobacter baumannii.
Next, the team turned to a massive digital database of about 1.5bn compounds to hunt for more new drugs. On 107 m of these, they set the algorithm to work. Three days later, the program returned a shortlist of 23 potential antibiotics, two of which seem especially potent. Now, the scientists intend to search the database for more. The conventional route of obtaining substances and testing them in the laboratory would have made it impossible to screen all 107 m compounds.
Now a fundamental question arises, with the recent outbreak of coronavirus, a large family of viruses that cause illness ranging from fever, common cold, cough, shortness of breath and breathing difficulties, to more severe infections that can cause pneumonia, severe acute respiratory syndrome, kidney failure, and even death. And the countries around the globe are stepping up efforts to tackle a new coronavirus that originated in China’s Wuhan city.
Can AI be key to coronavirus cure?
Responding to this question, Dr. Zubair Shah, Assistant Professor at College of Science and Engineering Hamad Bin Khalifa University (HBKU), said that if complete data of those infected were available to AI algorithms, it could reveal the factors that contributed to the recovery. AI may also be imperative in determining the most effective treatment plans.
“As of today, the total number of confirmed coronavirus cases amounts to 78,766, out of which 2,461 have died from the virus, and 23,133 have recovered. In answering the question of why some have managed to recover, artificial intelligence (AI) may be of decisive importance. If complete data of those infected were available to AI algorithms, it could reveal the factors that contributed to the recovery. AI may also be imperative in determining the most effective treatment plans. AI incorporating genetic, biological and environmental data of those infected, its algorithms may ultimately aid in the discovery of a cure for the coronavirus,” he added.
“The global biomedical community is constantly seeking and evaluating new medical countermeasures in its pursuit of a cure for the coronavirus. The disease could provide opportunities for in-silico screening of drugs as a way to determine their effectiveness, and decipher the structure of possible vaccines. In-silico screening involves the use of sophisticated computer modeling for medical analysis. But for this to be effective, extensive testing in the lab or in-vivo would be required,” said Dr. Hadi M. Yassine, an Associate Professor of infectious diseases and a Research Projects Manager at the Qatar University Biomedical Research Center (BRC). Hadi serves as an Adjunct Faculty at Hamad Bin Khalifa University in Doha.
