BURLINGAME, Calif. — Scientists have collaborated to develop a way to reproducibly find promising molecules against the bacteria tuberculosis (TB) in silico, using off the shelf software. A team led by Sean Ekins at Collaborative Drug Discovery in Burlingame, CA with co-authors at the University of Medicine and Dentistry of New Jersey, Southern Research Institute, and the Institute for Tuberculosis Research, University of Illinois at Chicago, has published a study in PLOS ONE entitled Enhancing Hit Identification in Mycobacterium tuberculosis Drug Discovery Using Validated Dual-Event Bayesian Models. This work is a follow up to a recent study in which the same team reported the first such combined model, able to sift through drug libraries and pick out compounds likely to target TB with minimal toxicity to humans, as reported in the journal Chemistry & Biology.
The current study demonstrates the largest prospective validation of such computational models to date, screening in silico 82,403 molecules with the models, of which 550 molecules were tested in the lab, finally identifying 124 molecules with promising activity against TB. Different sets of drug libraries were tested, with 15-28% found to be active. Normally, when very large compound libraries are screened, less than 1% are found to be active. In addition, a set of kinase inhibitors was screened, which contained several FDA approved and late stage clinical candidate compounds used in cancer treatment. Several of these had interesting activity against TB in vitro, which may represent a starting point for further research.
All of the computational models used in the study were based on public datasets available in the CDD Vault ™, and are being made publically accessible for fostering collaboration. “Several years of such computational studies overwhelmingly argue for their inclusion prior to large-scale random screening for compounds active against TB. This focuses resources on testing compounds more likely to be active and without side effects, saving money for more expensive clinical studies” remarked the lead author Sean Ekins Ph.D., D.Sc the VP Science at Collaborative Drug Discovery.
Such an approach clearly represents a useful way to screen for other compounds as potential antibacterials, an area of urgent need based on drug resistance, as well as potential biological threats posed by some bacteria.
CDD Vault™ (www.collaborativedrug.com) is a hosted database solution for secure management and sharing of chemical and biological data. It lets you intuitively organize chemical structures and biological study data and collaborate with internal or external partners through an easy to use web interface. CDD Vault hosts a unique collection o f publically available structure-activity relationship data, enabling researchers to mine their privately stored data together with a variety of scientific data providers.
Media Contacts: Barry Bunin, PhD, Collaborative Drug Discovery, (650) 204-3084, info@collaborativedrug.com