Winter 2017 Interactions Collaboration Highlight: Visual Analytics Team Makes Complex Genetic, Health Data More Digestible, page 4 Winter 2017 Interactions highlights the collaborative activities of the Mayo Clinic & Illinois Alliance for Technology-Based Healthcare. Inside Knowledge Engine for Genomics Gains Momentum ..... 2 Visual Analytics for Precision Medicine ........…............. 4 UPDATE: Biomarker Challenge Projects ….................... 6 Recent Alliance Publication Abstracts and Grants …... 8 SURF Impact Continues .....................................…........ 10 Fellowships for Technology-Based Healthcare .........… 12 On the cover: Colleen Bushell, PhD, (left) and Heidi Nelson, MD (right) have worked together for years on various collaborative Individualizing Medicine Conference …....................... 13 projects that utilize visual analytics of high dimensional health data for explorative research and predictive modeling for a wide range of health issues. Interactions is published by the Interdisciplinary Health Sciences Initiative (IHSI) at Illinois: Neal Cohen, Director, IHSI Sue Johnson, Director, IHSI Communications Ashley Lawrence, Interactions Managing Editor 901 W. University Avenue, Suite 101 Urbana, Illinois. 61801 www.healthinitiative.illinois.edu DESIGN: Jason Michael Bentley WRITERS: Margret Berg Miller, Claudia Lutz, Ashley Lawrence PHOTOGRAPHY: Kathryn Faith, L. Brian Stauffer, Ashley Lawrence, Mayo Clinic. Check out the new and improved mayoillinois.org for the latest Alliance news and events. Interactions Mayo Clinic & Illinois Alliance Steering Committee Mayo Clinic Scott Beck James Buntrock Bruce Horazdovsky Kiaran McGee Jason Pitzen Chris Schad Administrator, Center for Vice Chair, Information Associate Dean, Mayo Associate Professor, Operations Administrator Research Operations Individualized Medicine Management & Analytics Graduate School Medical Physics Manager Keith Stewart Caer Rohrer Vitek Emily Wampfler Eric Wieben Mathieu Wiepert Director, Center for Education Program Business Development Director, Medical Unit Head, Information Individualized Medicine Manager, Center for Manager, Mayo Clinic Genome Facility Technology Individualized Medicine Ventures Illinois Rashid Bashir Colleen Bushell Matthew Hudson Liudmila Mainzer Margret Berg Miller RK Narayanan Department Head, Senior Research Scientist, Professor of Bioinformatics, Senior Research Scientist, Research Development Associate Technology Bioengineering Applied Research Institute Crop Sciences National Center for Manager, Interdisciplinary Manager, Office of Supercomputing Applications Health Sciences Initiative Technology Management (ex officio) Gene Robinson Saurabh Sinha Ruth Franklin Sosnoff Derek Wildman Jeff Woods Director, Carl R. Woese Professor, Computer Assistant Director, Professor, Molecular & Professor, Kinesiology & Institute for Genomic Biology Science Interdisciplinary Health Integrative Physiology Community Health Sciences Initiative (ex officio) Winter 2017 1 KnowEnG team members from Illinois and Mayo Clinic at their 2016 annual meeting in Urbana. The Knowledge Engine for Genomics Gains Momentum By Claudia Lutz, Carl R. Woese Institute for Genomic Biology F or the individuals involved in KnowEnG, a Center of connections to related research efforts. Abel Bliss Professor Excellence established by an NIH Big Data to Knowledge of Engineering Jiawei Han, a program co-director for the (BD2K) Initiative award to the University of Illinois in center, praised the rapid progress of the center, which collaboration with the Mayo Clinic, recent activities have was founded in October 2014. Efforts at Illinois and the highlighted both how much has Mayo Clinic have focused on software development and KnowEnG is being designed already been accomplished in the the creation of resources for bioinformatics education and with a user-friendly interface past two years and the promise of training tailored to the needs of biomedical researchers, the center’s newest directions. as well as establishing a cycle of feedback and refinement and standardized analytical that will ensure the utility and accessibility of the resources pipelines that guide KnowEnG stands for Knowledge under development. Engine for Genomics, representing investigators who are less the center’s mission to develop “We need to bring computational genomic science familiar with computational analytical resources that will allow together with biomedical science, and this [KnowEnG] researchers to better interpret new is an excellent effort,” said Richard Weinshilboum, MD, aspects of bioinformatic genomic results by leveraging Mary Lou and John H. Dasburg Professor of Cancer analyses, and offers varying community knowledge of how Genomics and director of the center’s Pharmacogenomics genes interact with each other. Translational Program. Weinshilboum praised the levels of customization Advances toward this goal productivity of the partnership between Illinois and the were evident when members of Mayo Clinic within the center. depending on the user’s the center’s External Advisory expertise. Committee and representatives of Since the center’s inception, researchers have developed a NIH visited Illinois to attend a two-day conference at the novel resource, the Knowledge Network, which integrates Carl R. Woese Institute for Genomic Biology, which houses many different high-quality public genomic information the center. sources into a comprehensive theoretical landscape of gene functions and interactions. By placing their new Presentations by investigators highlighted research results into the context of the Knowledge Network, achievements, tools developed, plans for collaborator biomedical researchers can produce a high-powered review, refinement, and dissemination, and new analysis of new genomic and transcriptomic datasets. 2 Interactions Another innovation of the KnowEnG Center is the Knowledge Engine (also called KnowEnG), an analytical Applying Big Data to Myriad platform that takes advantage of algorithms that are used successfully in other data mining endeavors, including Health Issues, Through KnowEnG Google’s search functions, but have not been previously applied to interpretation of genomic data. KnowEnG By Ashley Lawrence enables researchers to include the Knowledge Network, as O well as more traditional resources, as part of their analytical ne of the main paradigms embraced by the KnowEnG workflow. Center is the concept of “knowledge-guided analysis,” in which researchers can analyze their own data in the context KnowEnG is being designed with a user-friendly of publicly available data. Primary sources include gene interface and standardized analytical pipelines that guide expression data sets, gene homology relationships, protein- investigators who are less familiar with computational protein and gene-gene interactions, gene ontologies, and aspects of bioinformatic analyses, and offers varying literature-based relationships. levels of customization depending on the user’s expertise. The center has contracted with Amazon Cloud so that Professor Saurabh Sinha says the eventual idea is that researchers can run analyses on the cloud, without scientists will visit the KnowEnG portal with their own requiring local resources, thereby increasing the tool’s datasets (in spreadsheets) and use KnowEnG to analyze usability and making it possible to run multiple variations their data in the light of this massive compendium or aspects of an analysis simultaneously. (Knowledge Network) of community-shared datasets. “Cloud-based computing can be very helpful,” said He gives specific examples of analyses already done, Professor of Computer Science and Willett Faculty Scholar emphasizing novel algorithms that have been developed Saurabh Sinha, who serves as the second program co- and applied to the discovery of mechanisms underlying director for the center. “Users would like to run those diverse phenotypes such as: interactions in parallel and be done with it, and a cloud • A technique based on diffusion component analysis computing network makes that happen very easily.” that identifies cancer pathways associated with drug response; KnowEnG’s user interface will also connect with an extensive online suite of bioinformatics training modules • An approach using “network smoothing” of gene that can improve understanding of both computational expression data and random walks with restart on the and biological aspects of genomic analyses. Founder Knowledge Network to better rank cytotoxicity-related Professor of Bioengineering and of Physics Jun Song genes; leads training activities of the center, which include a new • A probabilistic graphic model that integrates genotype, partnership with Fisk University to provide bioinformatics gene expression, and transcription factor-DNA education and research opportunities to minority binding data with drug response to identify regulatory undergraduate students. mechanisms of drug response variation across individuals; and An important focus of the next phase of work will be ensuring that KnowEnG is accessible to biomedical • Random walk-based methods for gene set researchers at all levels of computational skill. Partnership characterization, as an alternative to existing techniques with the Mayo Clinic, as well as bio-focused research such gene set enrichment analysis, using it to glean groups at Illinois, will provide opportunities for several systems-level insights about social regulation of rounds of user testing and feedback. After one more aggressive behavior. productive year, the KnowEnG Center is excited to share their progress with the wider world of genomic biomedical For more information on the KnowEng Center, including a research. list of publications, visit www.knoweng.org. Winter 2017 3 Visual Analytics for Precision Medicine Research Collaboration Highlight By Ashley Lawrence I t really all began thanks to Dr. Edison Liu’s sabbatical. At ‘Genetics for Dummies’—just so I could get the terminology least, this is how Colleen Bushell, senior research scientist at the correct,” says Bushell. “But even back in the early NCSA days, University of Illinois’ Applied Research Institute (ARI), explains I was always working in a space like physics or atmospheric her visual analytics team’s leap into the genomic and medical science, where it was always a collaboration and I was simply data domain. trying to understand enough to represent it accurately.” It was 2010 and Liu, president and CEO of The Jackson With the formalization of the Mayo Clinic and Illinois Alliance, Laboratory and adjunct professor of molecular and cellular Bushell, Welge, and other members of their NCSA visual biology at the University of Illinois, was in Urbana-Champaign analytics team—mathematicians, information designers, and on sabbatical and meeting with mathematicians, computer software developers—came together to focus their expertise scientists, and fellow biologists at the National Center for on health and medical projects. The ARI team now includes Supercomputing Applications (NCSA). As the leader of a world- Bushell, Loretta Auvil, Matt Berry, Lisa Gatzke, Peter Groves, class, cutting-edge genomic research institution, Liu admitted Xiaoxia Liao, and Michael Welge. a struggle: he always had trouble communicating the results of genetic analysis of a tumor to physicians. The information was Clearly Communicating Genetic Complexities so complex, he didn’t know where to start. The visual analytics pilot project, an early collaboration of the Mayo Clinic and Illinois Alliance, along with the Genome Michael Welge, long-time colleague of Bushell’s at NCSA and Institute of Singapore (Edison Liu, founder) was a clinomics then-director of data-intensive technologies and applications success. What started out as a few rough design concepts research, heard Liu’s challenge and knew it was not unique. of genetic analysis of a tumor grew into a prototype genetic Welge, an early member of the Mayo Clinic and Illinois report that focused on a comprehensive diagnostic panel Alliance, was aware that Mayo Clinic researchers and physicians of 17 hereditary colon cancer genes using next-generation were dealing with the same difficulty—how to draw out and sequencing technologies. Previously, Mayo Clinic ran individual communicate relevant information from a flood of genomic tests for five genes most frequently mutated in hereditary data. The challenge aligned with the Mayo Clinic Center for colorectal cancer cases. Labs around the world could test Individualized Medicine’s Clinomics Program aim: to transform additional genes that were known or suspected to play a genomic data into applications and information to guide role in the disease, but no other institution could test all 17 decisions that ultimately improve patient health care. Members comprehensively. of the newly formed Alliance proposed a pilot project, and Mayo Clinic provided seed funding for the work. Like that, the Matt Ferber, co-director of the Clinomics Program at Mayo, Alliance’s first visual analytics for precision medicine project enlisted the Illinois team to design a visual report for this genetic was born. colorectal cancer panel, one that helped communicate the The Gamut to Genetic Variances results of the panel to physicians. Bushell presented the report at the 2012 Individualizing Medicine Conference, sparking great Visual analytics has been a widely accepted methodology interest from physicians and investigators—especially colorectal for decades. Bushell describes it simply as the marriage of surgeon Heidi Nelson. information design and data science. It is an outgrowth of information visualization and scientific visualization, and has overlapping characteristics, but can attack problems whose size, complexity, and need for both human and machine analysis makes them otherwise unmanageable. While Welge brings an analytical and computational point of view to projects, Bushell’s formal training is in graphic design. Prior to her role with ARI, she was a professor in graphic design at the U of I. Her interest was information design, focusing on how to communicate both static and interactive data most effectively. From 1986 to 2004, Bushell was an NCSA research affiliate in data visualization. She designed the interface for Mosaic, the world’s first Web browser, worked on taking numeric data from simulations on a supercomputer to create a visual animation of a thunderstorm developing, and co-developed a visual programming interface for NCSA’s The old genetic results report’s chronological structure, acronym-laden, and data mining software.“We were working with big data long jargon-filled narrative made it difficult for physicians to locate answers and was awkward to walk patients through. The updated report, on the right, has before it was called ‘big data,’” jokes Bushell. “Until about six an “information hierarchy” structure with analysis for varying levels to seven years ago, my knowledge of microbiology, genetics, of scientific and medical understanding. and medicine was zero. Nothing. The first book I bought was 4 Interactions different list of most-important features than traditional analysis found—or than what researchers had predicted. “It’s actually good that we’re not all biologists, because we present the results without any preconceived bias,” says Bushell. “With this project, the Mayo Clinic researchers were very excited, because the results showed genes that mapped to three biological pathways they believe to be relevant to this valve disease.” Creating New Tools for Medical Discovery and Decision-Making Nelson is pleased with the number of projects now underway between Mayo Clinic and the Illinois visual analytics group. Expanded Visualization Projects Meet “The work we’re doing together is critically important, because Extended Random Forest Colleen and her colleagues are able to take large amounts After seeing the results of the Alliance’s visual analytics of data and work with it to create two or more levels of collaboration first-hand, Dr. Heidi Nelson, also a vice chair for visualization. This is more comprehensible to a physician or research in the Mayo Clinic Department of Surgery and an patient than what we can provide alone,” says Nelson. Alliance collaborator in the microbiome research area, knew it was just the beginning. Nelson sponsored three additional Currently, the Illinois team is working with Dr. Nelson to Alliance visual analytics seed projects, each with different goals develop a microbiome research and reporting tool. It contains for the team to achieve. data from several microbiome studies where the microbial DNA have been extracted from stool samples and analyzed to “Sometimes the goal is to help understand the biological identify bacteria that are relevant to various diseases. The tool complexity of what’s happening, or build a predictive model. uses visualization techniques to characterize the microbiome Sometimes it is to determine feasible courses of action, given community and displays important bacteria on an interactive the genetic analysis. Sometimes the goal is to create an phylogenetic tree. Eventually, doctors could use the software interactive visualization tool, too,” says Bushell. to compare an individual patient’s profile with database results. One of Nelson’s seed projects that Bushell and the visual Bushell says she is particularly interested in the microbiome analytics team at Illinois ARI has completed for Dr. Jordan Miller’s project on a personal level. Two of her three children have type heart disease research lab at Mayo Clinic was analyzing high 1 diabetes, and recent research shows a connection between dimensional mRNA and miRNA data from myxomatous mitral the autoimmune disease and the microbiome. She says her valve heart disease (or mitral valve prolapse). Essentially the team’s focus remains to find ways to identify relevant features weakening or degeneration of the valve’s connective tissue, it in data and communicate intricate information in a biological is the culprit of many a valve replacement surgery. However, context that helps people make decisions. Bushell sees Nelson despite progress in valve repair, lower mortality rates, imaging, as an incredibly visionary part of their visual analytics team, and less invasive approaches, far too many patients undergo especially in moving these projects forward. unnecessary valve replacement procedures. Additionally, performing open-heart surgery, only to find the mitral valve is “What she wants to do is not just design a report of what we not at significant risk for repair or replacement, is unsafe and can find now, but push ahead to begin thinking about what we costly. Mayo Clinic researchers wanted to search for features that could potentially report on in five years, when we have more could help identify degenerative heart valves. data,” says Bushell. “She’s asking, ‘what are the kinds of things that clinicians want to know about the microbiome?’ She is Traditional statistical analysis notes the top ten up- and down- engaging and funding these projects, knowing that there is regulated features. Welge constructed a newer analytical not too much clinically actionable information to report on approach to the mitral valve project, aptly named Extended yet. Dr. Nelson wants to get the dialogue going and move Random Forest (ERF) because it builds off the well-known people toward a vision. Our design helps push the vision. It Random Forest algorithm. encourages them to say ‘we need to do this’ and ‘here’s how we can do it better.’” The method constructs hundreds of thousands of decision trees that, together, help determine top features and specific “What an academic institution likes to do is find opportunities combinations and strengths that are most meaningful to a for their people to solve problems. And in health care, we disease or condition. The process is extended to include stability have those real-world problems. So it’s a perfect marriage of analysis, which is important when there are many features having people with great expertise at Illinois filling the gaps we but small sample sizes. In this case, the ERF method found a identify at the Mayo Clinic,” says Nelson. Winter 2017 5 Update: Biomarker Discovery Challenge Projects By Margret Berg Miller I n 2014, the Mayo Clinic and Illinois Alliance awarded seed funding to four collaborative research teams to address three grand challenges related to biomarker discovery. Two projects have reached their conclusion within the seed program, and the Mayo Clinic and Illinois investigators are now using the data generated to expand their research. proportions are possible. Currently, the mechanism of TBI reactivation is unknown, but reactivation from a latent infection is the most prevalent source of transmission. The WHO cites gaps in testing and reporting for TB as a major challenge of this top 10 global cause of death. The Mayo Clinic and Illinois research group has been working to develop three POC device platforms for LTBI. The first is a POC multiplexed cytokine-chemokine detecting technology. The device uses a silicon photonic micro ring resonator (MRR) assay to detect various immune biomarkers in antigen-activated blood samples. This novel MRR technology has been clinically validated at the Mayo Clinic, and there is currently a manuscript in process detailing this work. Challenge One: Detection of The second device in development utilizes Biomolecules from Body Fluids for personal glucose Early Detection of Disease meters for POC diagnosis of LTBI. This challenge was taken on by a research team comprised The investigators of Dr. Patricio Escalante, from the Division of Pulmonary and are in the processes Critical Care Medicine at Mayo Clinic; Yi Lu, professor of of further validation chemistry at Illinois; Rashid Bashir, Abel Bliss Professor and of this technology. department head of bioengineering at Illinois; and Ryan However, they believe Bailey, professor of chemistry at the University of Michigan this device has great (formerly Illinois). Their project, titled “Novel Biomarkers potential for rapid and Point-of-Care Methods for Latent Tuberculosis diagnosis of LTBI Infection” aims to develop point-of-care (POC) diagnostic in resource-limited technologies that detect immune biomarkers for high-risk settings. latent tuberculosis infection (LTBI) in a variety of patient populations and settings, improve diagnosis of LTBI, and The third technology is a POC microfluidics platform that individualize patient management. has proven successful for other diagnostic applications, such as early sepsis detection, but is being improved upon Tuberculosis is a bacterial infection caused by to reach the level of detection needed for LTBI. mycobacterium tuberculosis. According to the World Health Organization (WHO) in its 2016 Global Tuberculosis Report, The Mayo Clinic and Illinois team intends to continue one third of the world’s population is infected. In 2015 their work to further improve and validate all three LTBI alone, 1.8 million people died from tuberculosis, and in the diagnostic technologies. Currently, four papers detailing same year another 10.4 million new cases were estimated, various pilot studies of these technologies are in the worldwide. Add the complex tuberculosis lifecycle, process of publication. Additionally, the group has several specifically the reactivation of dormant or latent infection grant proposals in preparation to acquire external funding (LTBI), plus vaccine resistance—and fallout of epidemic for their continued research and technology development. 6 Interactions The gelatin hydrogel platform developed by the Harley lab is proving to be a versatile tool for studying glioma progression, efficient evaluation of cell processes, and treatment efficacy. The platform can also be adapted to evaluate the response of patient derived samples, making this a promising technology for rapid assessment of personalized treatment strategies for GBM. Prof. Harley, along with Dr. Sarkaria, D.r Ian Parney, from the Department of Neurosurgery at Mayo Clinic, and Steven George, professor of biomedical engineering at Washington University in St. Louis, were recently awarded R01 funding from the National Cancer Institute, for their project titled “Biomimetic hydrogel niches to study the malignant phenotype of glioblastoma multiforme.” This award will allow the team to continue their work over the next five years. This team of inter-institutional researchers has used the biomarker discovery seed program as a launching point to enhance and expand the degree, scope, and external Challenge Three: 3-D Cancer funding potential of their collaborative interactions. Tumor Chip-Avatars for “We have established a pipeline for transferring patient Personalized Drug Therapy derived xenograft cells with a wide variety of diagnostic Brendan Harley, associate professor of chemical and and therapeutic trajectory information, gathered from biomolecular engineering at Illinois, and Drs. Daniel Ma more than 75 glioblastoma multiforme patients,” said and Jann Sarkaria from the Department of Radiation Harley. Oncology at Mayo Clinic received seed funding to address this challenge. Their project, titled “Chip-based Additionally, Illinois professors Rohit Bhargava and H. Rex engineered tumor microenvironments for glioma therapy” Gaskins have joined the project team. Several publications demonstrated the use of engineered glioma biomaterials are in preparation and two NIH grants pending, in addition to find successful therapies for glioblastoma multiforme to the R01 award already received, to round out the (GBM), which is the most common, aggressive, and deadly research activity. To elarn more about the continued glioma form of brain cancer and is often resistant to current avatar project, visit the Harley lab website at therapeutic approaches. harleylab.org. Winter 2017 7 Abstracts of Recent Publications Mapping Lung Tumor Cell Drug Prior Oral Contraceptive Use Responses as a Function of in Ovarian Cancer Patients: Matrix Context and Genotype Assessing Associations with Using Cell Microarrays Overall and Progression-Free Survival Kaylan KB, Gentile SD, Milling LE, Bhinge KN, Kosari F, Underhill GH Jatoi A, Foster NR, Kalli KR, Vierkant RA, Zhang Z, Larson MC, Fridley B, Goode EL Interactions between epithelial tumor cells and components of their microenvironment influence carcinoma progression Previous studies have shown that oral contraceptive use and affect how the cells respond to therapy. However, reduces the risk of ovarian cancer. With this in mind, the complexity of the extracellular matrix (ECM) makes it the researchers decided to examine the effect of oral difficult to investigate these interactions. This study takes contraceptive use on overall and progression-free survival the ECM environment into account by using a high- of patients diagnosed with ovarian cancer. A retrospective throughput cell microarray-based approach to study the cohort study was carried out on ovarian cancer patients impact of defined combinations of ECM proteins on lung who were seen at the Mayo Clinic over a span of 13 tumor cell drug responses. The effects of 55 different ECM years. When taking confounding factors into account— environments on the responses of lung adenocarcinoma like age, smoking, or family history—the data showed cells to cancer-relevant small molecule drugs were that patients who used oral contraceptives at any point evaluated with this approach. The researchers also assessed prior to being diagnosed with ovarian cancer had better the expression of neuroendocrine transcription factor progression-free survival than patients who had never used ASCL1, since there is a neuroendocrine-like subtype of lung oral contraceptives. Overall survival did not appear to be cancer that co-expresses ASCL1 and RET and has been affected by oral contraceptives when other factors were associated with reduced patient survival, increased tumor included in the analysis. Though this study suggests that cell proliferation, and expression of anti-apoptotic proteins. prior oral contraceptive use is associated with improved The results suggest that co-expression of specific ECM clinical outcomes for ovarian cancer, further investigation proteins with known genetic drivers in lung adenocarcinoma is needed to understand how and why this occurs. may impact therapeutic efficacy. The researchers believe Understanding the mechanisms behind these data may this approach could be utilized along with integration lead to more effective therapeutic interventions for ovarian of clinical cell samples and genomics data to define the cancer patients. molecular mechanisms by which cell–matrix interactions drive drug resistance. 8 Interactions