RCAC, College of Agriculture partner to reshape agriculture research
Purdue’s Rosen Center for Advanced Computing (RCAC) has partnered with the College of Agriculture to develop cyberinfrastructure that better serves the computational and data management needs of agriculture researchers.
The College of Agriculture’s Ag Alumni Seed Phenotyping Facility (AAPF) provides researchers with a controlled environment in which to run experiments on plants and simulate conditions like drought and different nutrient mixes. Experiments run in the AAPF can take months or even years to complete, and researchers can collect extensive data without damaging the plants via three different specialized cameras. These cameras collect information about the plants in multiple wavelengths, creating huge amounts of data in the form of RGB, hyperspectral, and x-ray images.
The College of Agriculture IT group partnered with the AAPF facility, select researchers, and an external vendor to create a high-performance computing and flexible storage solution to address the significant compute and storage required to process these images.
“One of our main focuses is to find the right solution for our researchers, whether that means leveraging existing campus technology, which is the preference, or creating something new,” said Eric Biggs, Director of Technology for the College of Agriculture. “We were aware Geddes was in the planning stages, but the AAPF initially needed a faster solution.”
Geddes, RCAC’s Community Composable Platform, is optimized for composable, cloud-like workflows and has flexible storage and connectivity options.
“Once Geddes was implemented, we immediately started working with RCAC to compare our solution with theirs, where we learned their enterprise offering would be our best path forward long term,” says Biggs. “It was clear their flexible system and dedicated support would allow us to focus more on research-specific needs and less on the underlying technology. By partnering with RCAC using Geddes, we can help our researchers save time and money, putting processed data in their hands faster and shortening the time to discovery “
Geddes is a containerized environment that lets the user build mini-containers that do specific tasks, and combine them together like building blocks to develop a customized pipeline that best suits a researcher’s individual needs. The fundamental building blocks can then be re-used to create a different pipeline for someone with different needs.
Rather than trying to develop a one-size-fits-all solution, which was the initial focus when working with the external vendor, the Geddes environment allows Biggs and his team to build a custom solution for each researcher, without having to start from scratch each time.
The AAPF also uses RCAC’s powerful new Negishi cluster to process the images from the hyperspectral cameras, which can reveal plant nutrient content, stress reaction, and disease and insect infection. Micro-containers in the Geddes environment process that data, transfer it to the clusters using Globus, back it up in RCAC’s Fortress data archive and move it to Data Depot, where it can be analyzed, all in one streamlined workflow.
After the success achieved with the AAPF project, Biggs knew that RCAC was the right partner to turn to for the College of Agriculture’s precision dairy farming research.
Precision dairy refers to using technology to manage the health of animals and detect both individual disease and herd-level problems early. At Purdue, that means ingesting data from seven different vendors in various formats.
Again, the Geddes environment was the right fit, and working with the Ag Data Services group, they were able to build containers that take those seven types of data and clean them, process them and generate files in a user-friendly format.
“When we partnered with RCAC it enabled us to respond faster to researchers’ needs,” says Biggs. “We’re no longer trying to maintain our own hardware and having to purchase new hardware. We don’t need to be trying to duplicate what RCAC is already doing, and with Geddes we can do things that are specific to each researcher, while still being able to reuse what we’ve already built.”
"We built Geddes to provide users the flexibility to create and manage their own reproducible data analysis infrastructure,” says Erik Gough, lead computational scientist for RCAC. “We're excited to see how the College of Agriculture is using this innovative system to meet the needs of their researchers."
Geddes is supported by a $400,000 award from the National Science Foundation (NSF award number #2018926). Preston Smith, executive director of Research Computing, is the PI on the project and Gough is a co-PI.
To learn more about Geddes or other Research Computing resources, contact rcac-help@purdue.edu.