Image
UVA's Cancer-Data Science Alliance Sparks Discovery
The inaugural Cancer Research and Data Science Symposium, co-hosted by the University of Virginia School of Data Science and the UVA Comprehensive Cancer Center on Feb. 19, opened the door to a powerful interdisciplinary collaboration that has the potential to transform complex biological data into actionable clinical insights.
Stephen Turner, assistant dean for research in the School of Data Science, and Chongzhi Zang, director of computational genomics at the Cancer Center, organized the event, which featured a welcome by Cancer Center Director Thomas Loughran, several research presentations, and over a dozen flash talks.
In his opening speech, Loughran mentioned that significant funding opportunities exist, and that he wants to continue collaborating with the School of Data Science to fund innovative grants.
Zang mentioned that cancer data science has gone far beyond genomics and now includes increasingly complicated data science. “We are right at the time to address these challenging problems,” he said.
Research talks spanned multiple disciplines, from studying gene interactions to using machine learning and functional genomics to understand cancer progression.
Some presenters had already received funding through the pilot fund program last year and shared work they had completed. Others, like Turner, had newly funded research that was underway.
Turner is collaborating with Roger Abounader, professor of microbiology, immunology, and cancer biology to use machine learning and functional genomics to understand the function of transcribed ultra conserved regions (TUCR) in glioblastoma.
“As many of you know, glioblastoma is a common, lethal central nervous system malignancy. Even with maximal therapy — chemo, radiation, surgery — its median survival is just a little over a year,” he said.
Turner added that decades of research have focused on the protein-coated portion of the genome, which is a tiny part of the genome. The vast, unexplored portion of the genome is the focus of their project.
While there are tens of thousands of publications related to some segments of the genome, he mentioned that TUCRs only have around 70 related publications, none of which were related to glioblastoma before this project.
Turner described the goals of his project in more depth and then spoke to the broader impact of these pilot programs: to generate data that can eventually lead to more funding.
“We really want to try to look at this as a model for Cancer Center and School of Data Science collaboration,” Turner said.
“We really want to try to look at this as a model for Cancer Center and School of Data Science collaboration,” Turner said.
Aiyang Zhang, assistant professor of data science, is collaborating with Zang of the Cancer Center. They received funding last year through the pilot program for their project, Topological Data Analysis (TDA) Based Approach for 3D Genome Dynamics in Lung Cancer Progression.
“We are trying to find the robust features that can enable multiscale, shape-based differential analysis of chromatin-tracing data,” Zhang said. “Overall, we want to develop a topology-based framework to quantify multiscale 3D chromatin reorganization in lung cancer.”
Following the research presentations were over a dozen flash talks, where each presenter had three minutes to explain their research interests and pitch ideas to possible collaborators in attendance.
Presenters discussed a wide range of topics, including research on understanding mitochondrial dysfunction in cancer, using multimodal AI to determine the effectiveness of cancer treatment, and leveraging data science to predict how tumors progress.
Karsten Siller, associate professor of data science, discussed three collaborative projects he is involved in with colleagues from the Cancer Center in his flash talk. The projects involve building spatiotemporal maps across biological scales to better understand both normal development and disease states like cancer. "The push to develop universal tooling is really in service of that bigger goal: enabling faster, more reproducible analysis and discovery," Siller said.
Assistant professors of data science Nur Yildirim and Tom Hartvigsen recently received a Darden LaCross AI Institute award for a project that Yildirim discussed during her flash talk.
“Imagine you’re working on a multimodal AI system, and AI thinks it’s cancer, and you know it’s not cancer,” Yildirim said. “If the end user wants to change the workings of the model because they know it’s wrong — because these models make mistakes all the time — how do we let people do that in an efficient way?” she asked. She and Hartvigsen are looking for clinical collaborators to join them in creating human-centered, participatory AI systems as an answer to that question.
Shiraz Robinson II, an M.S. in Data Science student ambassador, attended the symposium with a few other MSDS students, some of whom are in the genomics track. Robinson, who was a premedical student at the University of Maryland before pivoting to data science and quantum computing, stresses the value of interdisciplinary collaboration.
“Regions of the UVA academic ecosystem should be working in parallel, not in isolation,” he said. “The School of Data Science embodies the mantra that we are a school without walls.”
After the presentations, Turner announced that the call for proposals was open for the next round of Cancer Center/School of Data Science pilot program funding. He reiterated the immense value in bringing domain and computational expertise together.
“The School of Data Science is really committed to building on what we started here today,” Turner said.
He encouraged attendees to network during the cocktail hour that followed the presentations. “If a talk kind of sparked an idea, that’s a collaboration waiting to happen.”
Image
Since the symposium, the pilot program has received applications from eight teams with researchers representing both the Cancer Center and the School of Data Science. Turner expects that the program will fund five teams with decisions likely being announced later in the spring.
“Sixteen investigators applying across eight teams tells me the appetite for cancer and data science collaboration at UVA is real,” he said. “That was the whole point of the symposium: to surface the problems cancer researchers are wrestling with and match them up with data scientists who want in.”
