Faster, smarter and cheaper drug discovery
- Date:
- March 21, 2013
- Source:
- University of Medicine and Dentistry of New Jersey (UMDNJ)
- Summary:
- Computers are now sifting through drug libraries to pick out compounds likely to clobber TB with minimal side effects to humans. Programmers have ‘taught’ the computers to understand which chemical features of a drug are associated with efficacy against TB and which are associated with toxicity to mammalian cells. The process may lead to much less trial and error in finding new therapies. The computers even rediscovered a compound reported 40 years ago to have anti-TB activity but since forgotten.
- Share:
Scientists have developed a way to pinpoint promising drugs to fight tuberculosis (TB) and other diseases without setting foot in the laboratory. A team led by Sean Ekins at Collaborative Drug Discovery in Burlingame, Cal., and Joel Freundlich at the University of Medicine and Dentistry of New Jersey-New Jersey Medical School has trained computers to sift through drug libraries and pick out compounds likely to clobber TB with minimal side effects to humans, as reported in the journal Chemistry & Biology.
In the last four decades, only one new drug has been approved to treat TB, which causes nearly two million deaths each year. The process of identifying and testing potential new drugs has been streamlined in recent years by using robots to simultaneously test thousands of compounds in a process called high-throughput screening (HTS). Still, HTS costs millions and has generated few promising anti-TB drug leads, and even the most promising compounds often turn out to be toxic to human cells.
So cheminformatics expert Ekins and chemist Freundlich decided to use computers to do the initial legwork. By drawing from publicly available HTS data for TB, they 'taught' computers to understand which chemical features of a drug are associated with efficacy against TB and which are associated with toxicity to mammalian cells. Once trained, the computer successfully picked out agents proven to kill the TB bacteria in culture and even rediscovered a compound reported 40 years ago to have anti-TB activity but since forgotten.
Their new work shows that they can predict effective molecules prospectively using commercially available computer software and published HTS data. "If we can pick and choose a small number of compounds to test rather than screening libraries of thousands of molecules, then it's cheaper and immediately brings the compounds of most interest to the forefront," says lead author Ekins. "The data are out there, and we want to encourage people to use them," adds Freundlich, expressing a sentiment that may resonate with researchers given the current crunch on funding dollars.
The University of Medicine and Dentistry of New Jersey (UMDNJ) is New Jersey's only health sciences university with more than 6,000 students on five campuses attending three medical schools, the State's only dental school, a graduate school of biomedical sciences, a school of health related professions, a school of nursing and New Jersey's only school of public health. UMDNJ operates University Hospital, a Level I Trauma Center in Newark, and University Behavioral HealthCare, which provides a continuum of healthcare services with multiple locations throughout the State.
Story Source:
Materials provided by University of Medicine and Dentistry of New Jersey (UMDNJ). Note: Content may be edited for style and length.
Journal Reference:
- Sean Ekins, Robert C. Reynolds, Hiyun Kim, Mi-Sun Koo, Marilyn Ekonomidis, Meliza Talaue, Steve D. Paget, Lisa K. Woolhiser, Anne J. Lenaerts, Barry A. Bunin, Nancy Connell, Joel S. Freundlich. Bayesian Models Leveraging Bioactivity and Cytotoxicity Information for Drug Discovery. Chemistry & Biology, 2013; 20 (3): 370 DOI: 10.1016/j.chembiol.2013.01.011
Cite This Page: