UAMS Researcher Develops Model for Predicting Drug Toxicity

By ChaseYavondaC

As a step in this direction, University of Arkansas for Medical Sciences (UAMS) researcher Grover P. Miller, Ph.D., and collaborators have published a manuscript in the journal ACS Central Science describing their development of a model that predicts the reactivity of molecules toward biological molecules as possible causes underlying drug toxicity.

“Modeling Reactivity to Biological Macromolecules with a Deep Multitask Network,” by Miller and collaborators from Washington University in St. Louis — S. Joshua Swamidass, M.D., Ph.D.; Tyler B. Hughes, Ph.D. candidate; and Na Le Dang, M.D., Ph.D. candidate — was published July 29.

Their research is available to read on the ACS Central Science website at no cost and the researchers have provided open access to their model at http://swami.wustl.edu/xenosite/p/reactivity.

Their model did significantly better than standard methods at predicting known indicators for toxic reactions between drug molecules and DNA and protein.

For the entire data set of 2,803 molecules, their approach yielded totals of 257 (9.2 percent) and 227 (8.1 percent) molecules predicted to be reactive with DNA and protein, respectively. This selectivity is important because those interactions would have been missed by standard reactivity screening experiments used in drug discovery and development.

Miller, associate professor of biochemistry and molecular biology in the UAMS College of Medicine, credited a National Institutes of Health grant for bringing together his expertise in metabolism and experimental research and Swamidass’ expertise in computational models and statistics for developing and validating models of metabolism.

“The development of safer drugs requires knowledge of the mechanisms underlying adverse drug reactions so that they can be avoided during patient care,” Miller said. “Our model marks a significant advancement in not only identifying problem drugs but how to avoid those processes in guiding the development of safer, more effective drugs.”

ACS Central Science is a publication of the American Chemical Society that publishes select, stringently reviewed reports on research in chemistry and in allied fields, wherein chemical approaches play a key role.


UAMS is the state's only health sciences university, with colleges of Medicine, Nursing, Pharmacy, Health Professions and Public Health; a graduate school; hospital; a main campus in Little Rock; a Northwest Arkansas regional campus in Fayetteville; a statewide network of regional campuses; and seven institutes: the Winthrop P. Rockefeller Cancer Institute, Jackson T. Stephens Spine & Neurosciences Institute, Harvey & Bernice Jones Eye Institute, Psychiatric Research Institute, Donald W. Reynolds Institute on Aging, Translational Research Institute and Institute for Digital Health & Innovation. UAMS includes UAMS Health, a statewide health system that encompasses all of UAMS' clinical enterprise including its hospital, regional clinics and clinics it operates or staffs in cooperation with other providers. UAMS is the only adult Level 1 trauma center in the state. U.S. News & World Report recognized UAMS Medical Center as a Best Hospital for 2021-22; ranked its ear, nose and throat program among the top 50 nationwide for the third year; and named five areas as high performing — colon cancer surgery, diabetes, hip replacement, knee replacement and stroke. Forbes magazine ranked UAMS as seventh in the nation on its Best Employers for Diversity list. UAMS also ranked in the top 30% nationwide on Forbes’ Best Employers for Women list and was the only Arkansas employer included. UAMS has 2,876 students, 898 medical residents and six dental residents. It is the state's largest public employer with more than 10,000 employees, including 1,200 physicians who provide care to patients at UAMS, its regional campuses, Arkansas Children's, the VA Medical Center and Baptist Health. Visit www.uams.edu or www.uamshealth.com. Find us on Facebook, Twitter, YouTube or Instagram.

###