Researchers Use Machine Learning to Expedite Drug Discovery Process

Dev Kapadia  23′

Figure 1: The diagram gives examples of bacterial components that antibodies can target along with ways that bacteria can adapt for resistance. Researchers at the Ben-Gurion University of the Negev specifically targeted the ribosome, which is responsible for protein synthesis, to inhibit the M. tuberculosis bacteria.
Source: Wikimedia Commons

Western countries are known not only for drug discovery but also the ability for citizens to purchase many of these drugs. In fact, in 2019, Europe and North America were estimated to account for 30.8% and 21.9% of global pharmaceutical spending, respectively.¹ While access to drugs, and antibiotics in particular, is a good thing, their overuse can lead to antimicrobial resistance, which is when the antibiotics kill off the normal, vulnerable bacteria that allows for bacteria resistant to the antibody to multiply and become the norm. This problem is serious enough that the World Health Organization has labeled antimicrobial resistance one of the biggest threats to world health today.²

Two aspects of antimicrobial resistance make it extremely dangerous to public health: risk of infection with no pharmaceutical agents to treat it and disincentive for pharmaceutical providers to produce antibiotics, the reason for the latter problem being that companies will not want to pour money into a drug that is bound to become ineffective soon after its release. High research and development costs make the development of such a drug impractical.³ This, coupled with inflation and shrinking insurance coverage leads to an increase in drug prices that puts drugs out of reach for many citizens, depressing demand and further decreasing incentive for drug development.⁴ To end this current trend, companies must find a way to speed up drug discovery and cut down on cost.

To help make antibiotic drug discovery profitable once again, researchers at the Ben-Gurion University of the Negev in Israel developed a new system of antibiotic production in late 2019.⁵ Using the system, companies can discover and design drugs quicker at a fraction of the cost. Current methods of drug discovery study each bacterium and identify a selective target site where antibiotics can be used to inhibit the bacterium’s function. This is a time-consuming process, taking lots of time and capital before drug discovery can happen.

The system designed by the researchers at the Ben-Gurion University expedited the drug discovery process for the M. tuberculosis bacterium by isolating and targeting a segment of the RNA that expresses the peptidyl transferase gene. Peptidyl transferase catalyzes the formation of peptide bonds to build proteins and is common in all strains of the bacterium, even the drug-resistant ones. Machine learning was then used in the system to predict two inhibitors to this particular ribosomal segment, which were found to be more effective than the current antibiotic drugs used against the bacterium.⁵

The system was tested only on M. tuberculosis in the initial study. However, Dr. Barak Akabayov, one of the researchers from the team at the Ben-Gurion University, noted that the system can be used to target elements of other strains of bacterium.⁶ The specific gene sequence that the system targets might vary from bacterium to bacterium, but the system can employ machine learning to quickly adapt to these sequence differences and produce inhibitors far more efficiently than current methods. At the moment, Dr. Akabayov and the team are currently looking for an industry partner in order to further develop the product and secure a patent for the technology. In the future, the team hopes to minimize the current antibiotic drug shortage by scaling the technology so that it can expedite the drug discovery process globally.⁶

Bibliography

1. Koronios, E. (2019, June). Global Pharmaceuticals & Medicine Manufacturing. Retrieved March 2, 2020, from https://my-ibisworld-com.dartmouth.idm.oclc.org/gl/en/industry/c1933-gl/products-and-markets
2. (2018, February 5). Antibiotic resistance. World Health Organization. Retrieved February 21, 2020, from https://www.who.int/news-room/fact-sheets/detail/antibiotic-resistance
3. Ventola, C. L. (2015). The antibiotic resistance crisis: part 1: causes and threats. Pharmacy and Therapeutics 40(4): 277-283. PMID: 25859123.
4. Gill, L. L. (2019, November 26). The Shocking Rise of Prescription Drug Prices. Consumer Reports. Retrieved February 23, 2020, from https://www.consumerreports.org/drug-prices/the-shocking-rise-of-prescription-drug-prices/
5. Tam, B et al. (2019, August 6). Discovery of small-molecule inhibitors targeting the ribosomal peptidyl transferase center (PTC) of M. tuberculosis. Chemical Science, 38. https://doi.org/10.1039/C9SC02520K
6. Rees, V. (2020, February 12). Researchers develop screening platform to detect new antibiotics. Drug Target Review. Retrieved February 21, 2020, from https://www.drugtargetreview.com/news/55974/researchers-develop-screening-platform-to-detect-new-antibiotics/