The University of California, San Francisco's Institute for Computational Health Sciences has received a $10 million gift to support “data recycling” investigations. The approach to medical research involves mining existing data to potentially uncover new uses for existing drugs and help improve clinical care.

The University of California (UC), San Francisco, has received a $10 million gift in support of its newly launched Institute for Computational Health Sciences. The gift from philanthropists Mark Zuckerberg and Priscilla Chan, MD, helps fund the recruitment of new faculty members to UCSF, as well as the work of the institute's director, Atul Butte, MD, PhD, a pediatrician and medical informatics expert known for his “data recycling” approach to medical research.

Butte's lab develops sophisticated computational tools that mine existing, publicly available data for new insights into disease processes and approved drug treatments. He's currently helping to create a data warehouse and analytic platform containing clinical data from all six medical centers across the UC system, which collectively house electronic records on more than 15 million patients.

“We are barely tapping into the potential uses of all the clinical data being entered into electronic health records,” says Butte. “This database allows us to study hundreds of thousands of patients at the system level and potentially get a better idea of what treatments are working or not working for different types of cancer.”

Although not a substitute for the traditional clinical research method of recruiting patients into clinical trials, analyzing existing data can be a more cost-effective approach, notes Martin J. Murphy, DMedSc, PhD, CEO of Project Data Sphere, an online database of clinical trial data freely available to registered users. Discoveries are possible without collecting any new information—if you have a large enough data set and know where and how to look.

“The UC system has very deep pockets of data, especially in the oncology area,” Murphy says. “With such a large denominator, it's possible to pick up tiny signals that would be hidden in a smaller database, such as expressions of efficacy or hidden toxicities associated with specific genomic profiles.”

Findings from a recent study led by Butte demonstrate how the data-mining approach can work. In a Nature Communicationspaper, his research team demonstrates that an approved drug—pyrvinium pamoate—used to treat pinworms can shrink hepatocellular carcinoma in mice. The work involved analyzing data from several large databases, including The Cancer Genome Atlas and the Library of Integrated Network-based Cellular Signatures.

As part of the study, the researchers developed the Reverse Gene Expression Score, a ranking system that predicts how a particular drug could reverse the gene-expression profile in a particular disease. The method has the potential to aid in screening drug candidates and identifying novel drug targets.

“Eventually, we hope that this kind of analysis will help oncologists better tailor treatments to individual patients,” says Butte. “A couple hundred thousand patients with cancer come to UC every year—it's a huge store of real-world evidence that we can learn from.” –Janet Colwell