In May of 2019, UITS Research Technologies (RT) held its inaugural Student Poster Expo, offering students from Indiana University’s seven campuses the opportunity to showcase their work for the university community during the Research Services Expo. Students from any discipline who use advanced cyberinfrastructure in their research were eligible, and four received awards.
“Best Overall Poster” was awarded to John Clere for the project entitled “Floyd County history now in 3D: Supercomputing and photogrammetry increase understanding of artifacts.” The poster depicts the work of students at the IU Southeast Library in New Albany, IN using photogrammetry - the process of turning hundreds of photos into an accurate 3D computer model - to digitize objects of historical significance within the Floyd County area. The team took hundreds of pictures of each object, and then processed the images using Karst, IU’s high-throughput computing cluster, and a program called Photoscan, which allowed the team to turn the photos of each object into models. The process takes an average of eight hours per object to digitize with Karst’s processing power; without it, each model would take hundreds of hours. These models make it possible for researchers to examine each object - even fragile ones - over and over from any location.
Alexandra Bailey, a student in the IU Kokomo School of Sciences, won “Best Use of High Performance Computing” for the poster, “Examining the Schönberg-Chandrasekhar instability.” The poster focuses on the Schönberg-Chandrasekhar (S-C) limit, the point at which a star becomes a red giant and the limit of how far that could expand. A red giant is a star that has high luminosity, but low surface temperature. Stars enter this stage late in their evolution when the hydrogen in their core is exhausted, which will ultimately happen for the Earth’s Sun when it is approximately ten billion years old. When the Sun becomes a red giant, it will expand from its current size to a size closer to the orbit of Earth, though the details for how a star can change its structure so drastically is still an open question. The Schönberg-Chandrasekhar (S-C) limit represents the maximum mass that a non-fusing, isothermal core can have and still support a stellar envelope. Bailey is particularly interested in examining how the Schönberg-Chandrasekhar instability develops. If the instability is dynamic, it can play a role in the rapid readjustment of stellar structure when a star becomes a red giant. Bailey and her coauthors, Nimra Afreen and Patrick M. Motl, ran their simulations on Big Red II.
Eliza Foran of the IU Bloomington Department of Biology and Nicholas Brunk of the IU Bloomington Department of Intelligent Systems Engineering tied for “Best Visualization.” Foran’s poster, “Developing a workflow for bioacoustic recording devices and frog call analysis within Jetstream,” which emerged out of a need for effective, digitized workflows on the part of field biologists and environmental surveyors. The National Center for Genome Analysis Support (NCGAS) previously built a field station workflow supported by the NSF-funded national research cloud Jetstream, which allows researchers to collect and access atmospheric data via preconfigured virtual machines. To further field station support, the team wanted to fulfill demand for bigger data and remote data pickup. With a focus on bioacoustics, Foran and the NCGAS team designed a flexible workflow allowing automated surveying, data pickup and storage, and analysis of local frog calls. Acoustic observation of frogs is generally accomplished by recruiting volunteers with call-identification training. Automation reduces environmental interference, human labor, and also enhances identification with proper bioacoustic analysis. Using cheap hardware and open-source software such as Raspberry Pi microcomputers and WarbleR, NCGAS has created an accessible field station cyberinfrastructure workflow for frog surveyors and field biologists. When all the data is collected, it is visualized on a handsome webpage as a principal component analysis plot, spectrogram, and lossless sound file for proofing. Automatic data collection and pickup allows researchers to focus on bioacoustic analysis rather than wandering the woods at night with a microphone.
Nicholas Brunk worked with faculty members Masaki Uchida, Trevor Douglas, and Vikram Jadhao on the poster “Self-assembling Ordered Arrays of Virus-like Particles Mediated by Linkers,” the results of which were recently published in ACS Applied Bio Materials. The team established a simulation-based theoretical foundation for experiments on the self-assembly of virus linker arrays using IU HPC resources, enabling an understanding of the biophysics underlying functional 3D materials generated out of biological building blocks (viruses). In collaboration with the Department of Chemistry, their experiments mixed virus shells, small linkers, and salt in various amounts in order to create different types of assemblies, like ordered lattices or disordered globs, or to prevent assembly entirely. To make their experiments realistic, the team created a simulated test tube with hundreds of viruses and hundreds of thousands of linkers in the presence of salt, with simulations correctly predicting the structure and capacity control of the arrays. The team faced the two-fold challenge of creating a model simple enough to be computationally feasible, yet realistic enough to be experimentally relevant, capable of making accurate predictions. The only way to model the system effectively was by using HPC resources. Even with simplifications, each of the team’s simulated test tubes used 64 processors and hundreds of hours of wall time on Big Red II and analysis required over a hundred simulations; collectively, they generated over a terabyte of data stored on the IU Data Capacitor and backed up on the Scholarly Data Archive and Box. The team has demonstrated that these virus arrays speed up reactions between cargo that may be packaged in the virus shell nanoreactors; this may prove useful in both industrial applications and academic labs.