PSC in Brief
PSC Projects Earn 2016 HPCwire Awards
Two PSC projects were cited in the annual HPCwire Readers’ and Editors’ Choice Awards in Salt Lake City, Utah, in November. PSC’s Bridges supercomputer won “Best Data-Intensive System (End User Focused),” and its ongoing collaboration to reconstruct the microscopic architecture of the brain with Harvard University and the Allen Institute for Brain Science won “Best Use of High Performance Data Analytics.”
HPCwire, the leading trade publication in the supercomputing field, bestowed the awards at the 2016 International Conference for High Performance Computing, Networking, Storage and Analysis (SC16).
Intensive Approach May Help Diversify Bioinformatics
Bringing more minority students into the training pipeline in bioinformatics—the use of computers to analyze biological data—may require a major rethink of STEM outreach to minority-serving institutions (MSIs) and more funding, a team from the University of Puerto Rico and PSC reported at the XSEDE16 conference in Miami in July 2016. The group surveyed students from MSIs before and after an intensive two-month summer internship in bioinformatics offered by PSC’s NIH-funded Minority Access to Research Careers (MARC) program.
While PSC’s summer internship raised participating students’ bioinformatics skills across the board, the students’ lack of previous exposure to basic bioinformatics skills was telling. Despite being among the top students at their institutions, between 59 and 91 percent ranked at the “novice” or “basic” level in a set of computer skills critical for careers in biology today. These skills are particularly important, because having some exposure to them is necessary to derive benefit from the one- to two-day bioinformatics outreach program typical in the field, according to authors Ricardo Gonzalez Mendez of the University of Puerto Rico, Medical Sciences, and PSC’s Pallavi Ishwad and Alexander Ropelewski.
Making Big Data DANCE(S)
A project to make movement of the largest datasets more efficient has reached a major milestone, according to researchers from PSC and their colleagues in the National Science Foundation’s XSEDE collaboration of supercomputing sites. For the first time, the DANCES project has tested networking hardware components necessary for scheduling network bandwidth using Software Defined Networking (SDN) between two national high-performance computing (HPC) sites.
The NSF-funded DANCES (Developing Applications with Networking Capabilities via End-to-End SDN) is researching a kind of on-demand high-occupancy vehicle lane for large data transfers across networks, allowing Big Data users to specify how much bandwidth they would like to use and schedule their job so that they can be assured the data will get through in a timely fashion, with minimal interference from and to other users. The researchers, including PSC’s Kathy Benninger, presented their results at the XSEDE16 conference in Miami in July 2016.
Galaxy Gateway Offers Transparent Access to PSC’s Bridges
Researchers using the free Galaxy scientific analysis gateway to assemble and identify the active genes in a given species now have transparent access to PSC’s Bridges supercomputer. The Pennsylvania State University-run Galaxy now allows massive assemblies of RNA sequences—the “translated” version of genes that are active in a cell—being performed by the Trinity software tool on Galaxy to run automatically on “large memory nodes” on Bridges, without the need for users to obtain individual allocations on the PSC system.
For more information about Galaxy and Bridges see https://galaxyproject.org/ and https://www.psc.edu/bridges.