The group tackled projects including improving data collection from unmanned aerial vehicles (UAVs), improving data and metadata management practices, and developing strategies for leveraging Open Science Grid (OSG), a National Science Foundation and Department of Energy project that aims to facilitate access to distributed high-throughput computing resources for U.S.-based researchers.
Metadata discussions centered on data and metadata structure as well as metadata terms, aiming to create streamlined methods for researchers everywhere to standardize metadata reporting and data access for future studies. A breakout group discussed creating an easy process for scientists to define how their data were collected, developing a map that links metadata terms across different vocabularies to facilitate cross-referencing, cataloging image data and metadata, and ensuring public datasets can be easily tagged with permanent identifiers for discoverability.
Future hackathons may focus on developing a more automated way to maintain and update metadata mappings as a reference for the global phenotyping community.
"I think getting this group together to solve some tractable problems was great," noted Carolyn Lawrence-Dill, an associate professor of genetics, development and cell biology at Iowa State University and hackathon attendee. "Not only did attendees articulate some problems in high-throughput phenotyping data acquisition, analysis, and dissemination, they made progress toward their resolution and began the groundwork toward establishing a community of practice."
Another breakout group worked to develop a solution for easy, rapid data quality assessment for UAVs, which are commonly used for phenotyping and crop measurement, so that the pilot knows whether the run produced good data or if the flight needs to be done again.
"Checking data that you've collected while you are out the field is usually stressful for technicians and scientists in the field, and it’s stressful for principal investigators and people waiting to get ‘good’ data,” said Blake Joyce, a CyVerse science informatician who participated in the hackathon. “And now that we can collect hundreds of gigabytes in the field from cameras on UAVs, that stress can be magnified hundreds of times.”
Joyce’s group developed a simple program to check images, UAV flight paths, and positioning to ensure high quality images are collected. They plan to continue working on the program to formalize it into a graphical interface.
Said Jennifer Clarke, Director of the Quanitative Life Sciences Initiative and an associate professor at the University of Nebraska-Lincoln, who attended the hackathon: "I was impressed by the progress that was made towards solving computational and data challenges in plant phenotyping. It was an excellent opportunity to learn about the capabilities at CyVerse, interact with experts from multiple institutions involved in phenotyping, and share knowledge and expertise."
Participants included researchers and informaticians from CyVerse, the UA, Arkansas State University, University of Arkansas, Iowa State University, University of Nebraska, University of Nebraska–Lincoln, USDA, Purdue University, University of California, Irvine, the Donald Danforth Plant Science Center, and the National Center for Supercomputing Applications at the University of Illinois.