Growing with HPC
Much of the HPC culture has to do with growth, whether expanding our predictions about the future of machinery or figuring out new ways to interact with particles. For Associate Professor Carolyn Lawrence, HPC helps her work with natural growth - in ways that were never before possible.
Lawrence's work involves phenotypic prediction in plants. Phenotypes are the observable characteristics of living things including plants. Color, plant height, and response to stress are a few examples. Phenotype is important to researchers like Lawrence because the ability to manipulate phenotype allows them to make plants more productive in diverse situations.
Lawrence is focusing on predictive phenomics, or figuring out how a plant would develop or a gene would function in an environment that it has not been grown in before.
"People used to say that you couldn't predict that sort of thing. But by extrapolating from existing data, you can predict what plants are likely to grow like in the new situation," Lawrence said. "Not only can you select existing genotypes to grow in existing environments, you can create genotypes that are likely to do better in a desired or anticipated future environment."
The impact is massive. Because of past advancements to modulate phenotypes, plants like corn have been gradually altered to yield much more product per acre over time. Two characteristics that have been changed in modern day corn are the angle of the leaves on each stalk and tolerance by root systems to growing closely together. A century ago, field corn could not have been grown at the high densities that it is today, due to the horizontal angle of the leaves and sensitivity of the root system to crowding. By altering plant architecture and selecting for plants that are less sensitive to crowding, scientists were able to supply to farmers with plants that yield much more product per acre overall.
"We're planting these seeds much closer together than ever before, and they can handle it," Lawrence said. "We also saw benefits from this research when the last big drought hit. Many people wondered how their crops did as well as they did. Farmers, companies, and universities like us pointed at the fact that improved genetics as well as improved management practices had made the crop more resilient. The idea of predictive phenomics is to plan for such conditions rather than relying on serendipity."
The predictions that HPC makes in regards to these phenotypes are largely based on analyzing data that have already been collected. Still, Lawrence said that the time it takes to complete a project can be anywhere from minutes to weeks.
"We're really lucky at ISU because we have such good access to data sets and repositories that already exist; our networking and computational infrastructure are really well planned out. The future is being able to access these types of information and compute them using the same machines," Lawrence said.
Lawrence is thrilled to be able to be in Ames and said that Iowa State is the best place in the world to be working on improving crops using computer science.