Data Science Collaborations
THE OPPORTUNITY: The Cascadia region of North America is home to some of the world’s leading technology, research, and medical organizations. Scientific collaboration across these research institutions will accelerate research — at the center of this collaboration is a focus on the coordinated generation of data and effective sharing. Development of robust regional collaborations and a data sharing ecosystem will position the Pacific Northwest as a global innovator in biomedical research and healthcare, now and into the future.
THE CHALLENGE: Cultural, technical and policy barriers have historically limited cross-organizational data sharing. The Cascadia Data Alliance will change that.
The Cascadia Data Alliance created the Cascadia Data Discovery Initiative (CCD) to spur regional cross-institutional collaboration. As a result of CDDI’s efforts, Cascadia has funded pilot projects and partnered with the Specimen Acquisition Data Network to accelerate data sharing.
Regional Collaboration
Three pilot projects have received funding that: (1) answer an important scientific question that benefits from regional collaboration; and (2) develop or test innovative approaches to using technical solutions, data/methods standardization, best practices, and/or cloud services (using Microsoft Azure) that can be generalized to future collaborative efforts. These projects are made possible thanks to financial and in-kind support from Microsoft and institutional support from each participating organization.
Cloud-Based Genome Analysis for Monitoring Breast Cancer
A breast tumor’s genetics can change as the cancer progresses throughout treatment. But the changes don’t happen universally throughout all cells in the tumor, leading to distinct populations of cancer cells. If some of the cell populations have mutations that confer drug resistance, this can become a serious problem for the patient. In this new project, the research team will use state-of-the-art genomic methods to determine how tumor cells’ genetic profiles change during treatment and how those changes are linked to the patient’s outcomes.
Computational biologist Dr. Gavin Ha of Fred Hutch is one of the project’s leaders, along with BC Cancer’s Drs. Andrew Roth and Samuel Aparicio, and Dr. Natasha Hunter of the University of Washington Medicine and Fred Hutch.
How the Gut Microbiome Affects Cancer Immunotherapy Drugs
Drugs called checkpoint inhibitors that release the power of the immune system against tumors are important treatments for many different types of cancer. But the drugs’ effectiveness and side effects vary from patient to patient — and research has shown that the communities of bacteria in a patient’s body, or microbiome, plays a role. But which bacterial species are most important? A team of scientists hopes to find out. They will create a large, multi-institutional repository of samples of mouth tissues and stool from people who are using checkpoint inhibitors for cancer treatment.
The researchers — led by Fred Hutch’s Dr. David Fredricks, BC Cancer’s Dr. Kerry Savage and OHSU’s Dr. Morgan Hakki — will harness cutting-edge genomic technologies, data analysis tools and cloud computing to reveal the genetic fingerprints of the bacteria in the gut and match them to the outcomes of the patients’ treatment.
Using Machine Learning to Help Diagnose Specific Cancer Types
As targeted treatments have become available for specific types of ovarian cancer, it’s become more important than ever that each patient receives an accurate diagnosis of her tumor type. To help make this possible, a research team wants to establish an international network for AI-based, privacy-protected pathology quality assurance. Ovarian cancer will be the team’s proof of concept for a system that could eventually be used for a variety of cancers.
The team is led by Fred Hutch’s Dr. Holly Harris, BC Cancer’s Dr. David Huntsman, OHSU’s Dr. Terry Morgan and Dr. Ivan Beschastnikh, a computer scientist at UBC. It also includes Simon Fraser University's Dr. Tania Bubela and UBC's Dr. Ali Bashashati, who developed the original algorithm the group hopes to generalize.
Microsoft Azure Cloud Credits
The Cascadia Data Alliance recently made Microsoft Azure computing credits available for interdisciplinary regional collaborations between investigators at Fred Hutch, BC Cancer, UBC Data Science Institute, UW e-Science Institute and/or OHSU Knight Cancer Institute. Scientific projects involving basic, clinical or public health research were equally considered, focusing on validating new computational tools or building upon existing tools.
Applications were due February 4, 2022, and Cascadia Collaboration Awards have now been announced.
The Microsoft Azure computing credits will go to two projects:
1. Testing privacy and utility of 3D synthetic data generation and downstream models
Principal Investigator:
- Dr. Jean-Francois Rajotte, Research Associate, University of British Columbia (UBC) Data Science Institute
Co-Investigators:
- Dr. Juan M. Lavista Ferres, Chief Scientist and Lab Director, Microsoft AI for Good Research Lab
- Dr. Arman Rahmim, Senior Scientist and Professor, BC Cancer/UBC
This team will be testing the privacy of 3-D synthetic medical image generation that can train downstream models. The team will use a generation method that they recently developed, and use known attack formats, including ones adapted from 2-D implementations. To address the privacy-utility trade-off, they will define utility as the performance of a downstream task, such as classification and segmentation.
2. Eavesdropping on communications between cancer and immune cells
Principal Investigator:
- Yongjin Park, Scientist and Assistant Professor, BC Cancer/UBC
Collaborators:
- Dr. Samuel Aparicio, Distinguished Scientist, Professor and Department Head, BC Cancer/UBC
- Dr. Ramon Klein Geltink, Assistant Professor, BC Cancer/UBC, and Investigator, BC Children's Hospital
- Dr. Poul Sorensen, Distinguished Scientist, Professor and Department Head, BC Cancer/UBC
This team hypothesizes that novel therapeutic targets in tumor-related immune and signaling pathways will be exposed and better understood by using single-cell RNA-sequencing data to study millions of cell-cell interaction patterns. Their long-term goal is to establish a novel framework of machine learning and Intelligence Augmentation to reveal the communications between cancer and immune cells, thereby advancing precision oncology.
Data Sharing
CDDI partners and the Specimen and Data Access Network (SAN) created a shared data and material use agreement. The Umbrella Agreement provides a streamlined process and easy-to-use templates to accelerate sharing of data and specimens across signatory organizations in the Pacific Northwest.
Preserving Privacy to Enable Data Access
Cascadia is also working to test data sharing methods that contain oftentimes sensitive information in a way that provides strong privacy preservation. Cascadia partners are working together to develop and demonstrate technology solutions that both preserve privacy and facilitate meaningful analysis of health research data.
Benefits
- Resources for Cascadia regional collaboration
- Closer ties to regional collaborators with complementary skills and expertise (and access to their broader networks)
- Reduction in red tape to accelerate data sharing