TURNING DATA INTO KNOWLEDGE FOR PRECISION MEDICINE

From left: Rahil Sethi, Jiefei Wang, Vishal Soman, Paul Cantalupo, Uma Chandran

Genomics—an interdisciplinary field bringing together health sciences and data science to study the structures and functions of an organism’s complete set of genetic material—is uniquely suited to the talents of researchers at Pitt. In studying the origins and course of breast cancer, blindness, dental deformities and dozens of other conditions, Pitt researchers are helping illuminate the ever-increasing complexity of the interplay of genes within humans, both traits that are shared and those that may be unique to one individual.

Built on the foundations of genetics and spurred by the achievement of the 2003 Human Genome Project, genomics is a field that has experienced frequent and ongoing revolutions.

Uma Chandran, research professor and director of the Pitt’s Genomics Analysis Core, explains.

“In the past you would cut a tumor into little pieces and sequence it and figure out what genes were expressed, or which genes were higher or lower compared to normal tissue.”

But a tumor consists of a mixture of cell types. As of early 2010, it was not possible to easily discover the variety of cells and microenvironment of a tumor, only whether it is expressing or not expressing particular genes. 

Fast forward to 2020. A researcher could now infer cell types in that tumor using new bioinformatics algorithms. Fast forward again to 2023. It is now possible to map the spatial architecture of the tumor and to answer questions such as the locations of immune cells, which can alter the effectiveness of therapies. 

“These advances are important because every person’s cancer is different, even if it’s the same type of cancer,” says Chandran. “That’s where personalized medicine comes in—we’re able to characterize different tissues in three-dimensional detail. Using that map of the tumor, it may be possible to predict how a particular person would respond to therapy.”

Incorporating machine learning and AI in the analysis tools makes these genomics tools even more powerful, and the demand for computing resources and trained workforce continues to grow exponentially. But the ability to create new data presents the conundrum familiar in data science—you now have more data than can be easily analyzed. 

“I often say we’re trying to turn data into knowledge,” says Adrian Lee, director of the Institute for Precision Medicine (a joint effort between Pitt and UPMC), and professor of pharmacology and chemical biology in the School of Medicine.

As part of making that knowledge available to researchers across the country, Pitt is now the data coordinating center for a large national program sequencing metastatic breast cancer for the Global Data Hub, led by Jonathan Silverstein, associate director of research informatics in the Institute of Precision Medicine and a professor of bioinformatics in the School of Medicine. The project was created by the Breast Cancer Research Foundation, which cited Pitt as pioneers in data coordination and sharing, and the University’s “vast experience coupling clinical and biological repositories with cutting-edge, secure high-performance computing.” Reflecting the unique possibilities of collaboration at Pitt, the project will also be able to call on the resources of the Pittsburgh Supercomputing Center (a center shared by Pitt and Carnegie Mellon University).

Chandran explains Pitt’s role in the Global Data Hub. 

“We host terabytes of data (a terabyte is 1000 gigabytes), integrating data from many sources and linking researchers to de-identified clinical data on patients. A significant part of the job is cleaning up the data for sharing with other institutions.”

The Global Data Hub creates larger sample sizes and more powerful studies. Pitt’s expertise in genomics allows the teams to understand the inevitable challenges, mistakes and errors, as well as vital issues of annotation, data provenance, data sharing and data security.

Lee believes that genomics has the potential to make profound impacts on medicine. “Take a very simple example in pharmacogenomics. If we both take the same drug, we will respond differently because of our genetics. If I give you Tylenol and I take Tylenol, we will metabolize it differently. That’s important for many drugs.”

For highly toxic drugs like cancer drugs, metabolizing the drug more slowly can be dangerous. Lee cites the known property of the blood thinner clopidogrel—a drug that is inactive by itself but plays a role in metabolizing an active drug. But 30% of individuals cannot metabolize clopidogrel.

“So you are not helping them,” Lee says. “If a drug is ineffective for an individual patient, or even dangerous, wouldn’t you want to know that up front? That is the kind of goal we hope for in precision medicine.”