Minute Marvels are really short videos depicting scholarship that Duke researchers have conducted with Duke Research Computing resources.
We’ll post a new Minute Marvel periodically, and we’re always looking for refreshing and innovative research applications of research computing technology.
Do you have a story? Let us know in an email to firstname.lastname@example.org! Send a brief description of what you’re up to and how you’re using Duke Research Computing to do your marvels.
The Minute Marvels
Ryan Baumann of the Duke Collaboratory for Classics Computing uses the Duke Compute Cluster to digitize thousands of Latin texts, dating back to the 1500s in some cases. His work one of the latest “classics computing” projects taking place a Duke, which has a digital humanities tradition going back the the 1980s. (Video production: Michael Blair, Devon Henry, Jeannine Sato, and Mark DeLong)
Dr. de March studies smells — how we detect them and the molecular “machines” that accomplish the work of disclosing odors to our consciousness. Her research in Duke’s Department of Molecular Genetics and Microbiology uses mouse models, cell culture, and GPU computing to learn the structure of proteins involved in a capturing tiny “odorant molecules” that make up the smallest units of smell. Dr. de March used Duke Research Computing’s GPU compute cluster to visualize the activity of the odorant receptor. The cluster consists of 16 Nvidia K80 GPU housed in four Dell C4130 computers.
Additional links: Nobel Prize in Physiology or Medicine (2004) announcement on “odorant receptors and the organization of the olfactory system” awarded to Richard Axel and Linda B. Buck. The Hiroaki Matsunami lab (https://mgm.duke.edu/faculty-and-research/primary-faculty/hiroaki-matsunami-phd/), where Dr. de March serves as a postdoctoral researcher.
“Humongous Fern-tastic Genomes” with Fay-Wei Li, PhD (August 2016)
Dr. Li studies ferns – “they’re fern-tastic!” He and his colleagues have assembled the genome of Adiantum shastense, a “maidenhair fern” from the area around Shasta Lake, California (hence, the name), which is currently the largest assembled fern genome. Li worked with genome sequence applications and used the Platanus Genome Assembler in particular. Raw data amounted to nearly a terabyte of Illumina sequence reads. Duke Research Computing supplied a large Cisco Systems UCS server with 2 terabytes of random access memory (RAM) and 72 CPU-cores (aka “Monster Kick-Ass Machine”) with seven terabytes of network attached storage from the Duke Data Commons, a large storage system with 1.5 petabytes of capacity made possible by a grant from the National Institutes of Health (1S10OD018164-01).
Dr. Fay-Wei Li is a postdoctoral associate in the Kathleen Pryer lab at Duke’s Department of Biology. Andy Ingham and Tom Milledge (Duke Research Computing) provided technical support for the machine and storage used in the project. (Video production: Jeannine Sato, Devon Henry, Michael Blair, and Mark DeLong).
Additional links: “Adiantum shastense, a new species of maidenhair fern from California” (PhytoKeys) (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4547024/), Kathleen Pryer lab website (http://pryerlab.biology.duke.edu/)
“3D Models, GPUs, & Duke Chapel” with Ed Triplett, PhD (June 2016)
Dr. Triplett and a group of staff, faculty, and students converged on Duke University Chapel and systematically photographed its newly renovated interior. The two dimensional photographs formed the basis of a three dimensional model produced using Agisoft PhotoScan software on Duke Research Computing’s new GPUs. The process of rendering 3D from 2D is called “photogrammetry.” The hardware provided by Duke Research Computing was a Dell C4130 computer containing four Nvidia K80 GPUs, running the Windows Server 2012 operating system. The project was generated by the Wired! Lab, a group of faculty committed to training students (and colleagues) in emerging digital visualization technologies for the study of historical sites, works of art, and urban environments (see the Wired! website: http://www.dukewired.org/).
Dr. Ed Triplett is CLIR/Duke University Postdoctoral Fellow in Data Curation for Visual Studies. Brian R. Norberg (Trinity Technology Services) provided technical support for the photogrammetry project. (Video production: Jeannine Sato, Michael Blair, and Mark DeLong; thanks to Duke’s Office of News and Communications for additional video)