Field Phenomics Workshop report (KSU, USDA)
Field Phenomics Workshop Report
Summary
This was the second Field Phenomics workshop organized by Jesse Poland and Kevin Wang at Kansas State University together with their USDA collaborators (Jeff White and possibly others). Both workshops took place at the University of Arizona Maricopa Agricultural Center. Ramona Walls participated last year. Martha Narro participated this year on the last two days of the 4-day workshop. I think iPlant may have been more involved in organizing the workshop last year. We weren't involved in organizing it this year. The value to iPlant is that this is a good workshop for connecting with users in need of support for managing very large datasets.
The workshop advertised nationally and, largely due to the success of the first workshop, this year's was oversubscribed. About 125 people applied for what initially was 30 participant openings, but was expanded to accommodate 50 participants. Participants were roughly comprised of 20 PI's level, 20 postdoc and graduate students, and 10 other technical staff. Most instructors/speakers were from academia, but there was also good representation from agricultural companies and sensor makers.
The workshop covered experimental design, sensor instrumentation, field vehicles and sensor integration data collection, data cleanup and quality assessment, data management, and data analysis. A lot of good sensors and methods exist, however all agreed that everything in field-based phenotyping is still a work in progress. There are no turnkey solutions. After defining the research goals and developing the experimental design, it's a process of trying different sensors and analysis methods to determine what works best for the particular crop, field and experimental situation.
Martha presented a 15 minute iPlant overview and 20 minute demos of the DE (data upload, workflow creation, running and monitoring analyses, looking at outputs), iDrop for data upload and a tiny bit on Bisque. Information and links on all other iPlant tools and services were included in the slides.
Bottlenecks
- Data management since a single sensor can generate 100's of GB of data per day during the growing season, and multiple sensors are used. All scientists are wrestling with how to manage this scale of data.
- Geospatial data quality assessment. After some initial pre-processing, the large amounts of data need to be spatially validated to ensure they are in the expected field plot, row and location.
Follow-up
- A call is scheduled for March 25th to discuss the computational support iPlant could provide for Field-based Phenotyping starting with what Jesse would like to do to make his workflow available to the community. The goal is to come up with a preliminary plan and identify next steps.
- Jorge Berny, graduate student at UC Davis, would like a letter of collaboration for a grant he plans to write. He has Martha's email address and will contact her when the time comes.
- Tami Stubbs, postdoc at Washington State University, is trying Bisque for image management. Martha and Ramona will follow up to see if she has questions, especially regarding metadata templates.
- As iPlant plans to support field phenomics data, (Martha thinks) we need to identify a group, or a few of them, who will really engage to work with us. We need to be prepared to handle the volume of data field phenotyping will generate. The group with the most momentum is Genomes to Fields. I'll have a better feel for how the Field Phenomics group is likely to progress after Wednesday's call. Martha will coordinate with Nicole to gather people for an internal discussion.