Artificial Intelligence Drug Discovery
Artificial Intelligence Drug Discovery : In recent years, many have expressed concern regarding the lack of new antibiotics being produced, as well as the compounding issue of antibiotic resistance. An increasing number of pathogens are becoming resistant to the antibiotics currently utilized.
With the high expense, considerable time investment, as well as an inability to target a wide range of chemical compounds, current methods of screening new antibiotics has been challenging.
However, researchers at MIT may have found a solution to the problem. Using a machine learning algorithm, researchers have been able to identify certain molecules that have characteristics that are highly desirable for an antibiotic. The computer model created is able to select certain molecules that have the optimal bacteria-killing potential that differs from pre-existing antibiotics.
This is accomplished through “in silico” screening, a method that analyzes large data sets and finds patterns to assess possible properties of a chemical compound. While “in-silico” screening has been previously available, it has not been accurate enough to redefine the drug discovery industry until now.
Researchers have used their robust algorithm to identify a specific compound that has the opportunity to revitalize the antibiotic industry. This compound is known as halicin, named after the artificial intelligence system in “2001: A Space Odyssey.” In clinical trials with mice, halicin was capable of killing many disease-causing bacteria, including several that are resistant to all currently known antibiotics, including Clostridium difficile, Acinetobacter baumannii, and Mycobacterium tuberculosis. Strains of E. coli were unable to develop resistance to halicin within a 30 day treatment period, whereas resistance developed to another antibiotic, ciprofloxacin, within 1 to 3 days.
Successful halicin trials have encouraged researchers at MIT to expand their computer model to other molecules. After screening more than 100 million molecules in only 3 days, the algorithm successfully identified 23 additional molecules that differed from current antibiotics structurally. Of the 23 candidates, 8 ultimately showed promise as an effective antibiotic.
Researchers hope to continue to utilize artificial intelligence in order to design new antibiotics, as well as to optimize pre-existing drugs. A possible area of future development includes implementing models that screen for antibiotics that target harmful bacteria without destroying beneficial bacteria in humans.
This combination of artificial intelligence and science has led to breakthroughs in antibiotic discovery and production. With future developments, the landscape of treatment of infectious diseases will continue to transform for years to come.