You are here
Mesabi, the Minnesota Supercomputing Institute’s newest supercomputer, was formally dedicated on April 23, 2015. Mesabi is the latest in a long line of supercomputers that have been serving the University of Minnesota research community for 30 years.
Mesabi is a heterogeneous compute system built by HP. Heterogeneous systems incorporate more than one type of processor. This increases performance by allowing specialized processors to handle specific tasks. Mesabi features an HP Apollo 6000 system with Intel Haswell processors and solid-state drives (SSDs), Mellanox Extended Data Rate (EDR) switching, and NVIDIA Graphics Processing Units (GPUs). MSI has also purchased 1 PB of Panasas storage for the system.
Mesabi by the numbers:
• Rated at 352 kW of power (enough to power 270 houses).
• 1.2 million BTUs required for cooling.
• Seven times as powerful as MSI’s Itasca supercomputer.
• More than 62 TB of memory.
• 480 Gbits/second connectivity to Panasas storage.
During the recent MSI Research Exhibition, several MSI users from diverse fields presented work that they will be transferring to Mesabi. Some examples are below:
Benjamin Brummel is a graduate student in the research group of Jonathan Sachs (Biomedical Engineering). Among other projects, the Sachs group is studying α-Synuclein, a protein found in the human brain that is associated with Parkinson’s disease. Mr. Brummel presented a poster titled “Lipids with one saturated and one polyunsaturated DHA chain homogenize lipid rafts and affect the membrane environment of α-Synuclein” (Benjamin E. Brummel, Anthony R. Braun, Jonathan N. Sachs).
Christine Dunbar is a post-doc in MSI Fellow Christopher Cramer’s research group. Professor Cramer is in the Department of Chemistry and is also a member of the Center for Sustainable Polymers and the Chemical Theory Center. Dr. Dunbar presented a poster describing research into sustainable alternatives to petroleum-based polymers (Christine R. Dunbar and Christopher J. Cramer. “Quantum chemical analysis of the mechanisms of epoxide/anhydride copolymerization using metal catalysts.”) The Cramer group develops and applies computational models to study chemical structures, properties, and reactivities in areas of chemical, biological, and environmental interest.
Chris Tessum is a post-doc in the group of Julian Marshall (Civil, Environmental, and Geo- Engineering; Fellow, Institute on the Environment) and Jason Hill (Bioproducts and Biosystems Engineering; Fellow, Institute on the Environment). This group uses models on the supercomputers to study the impacts on air pollution of different types of alternative fuels. They correlate changes made in the transport sector to air quality and public health. Dr. Tessum presented a poster describing how future use of conventional and alternative fuels may impact inequality and injustice in health effects related to air pollution (Christopher Tessum, Jason Hill, Julian Marshall. “Environmental justice and equality aspects of conventional and alternative light-duty transportation in the United States.”).
The name Mesabi was selected by MSI staff from over 140 suggestions from MSI users and others at the U. The Mesabi Range is the largest of the four major iron deposits in northern Minnesota that collectively make up the area known as the Iron Range. It is the chief deposit of iron ore in the U.S. The name comes from an Ojibwe word meaning “immense mountain.” The name reflects Minnesota’s cultural heritage and natural resources and ties in to an informal term for supercomputers, “Big Iron.”
posted on May 13, 2015
Research has shown that the human microbiome - the community of microorganisms that inhabit our bodies - may have health implications. It’s possible, for example, that a certain mixture of these microorganisms may cause certain diseases.
Researchers have recently begun to study the microbiome in depth. While much recent research has involved the bacteria that live in human bodies, only limited studies have dealt with the fungi that are also present. While there are fewer fungi present in the microbiome than there are bacteria, fungi also affect human health. For example, fungi cause some life-threatening diseases in infants, and other fungi may alleviate intestinal inflammation.
In a recent paper in PLoS One, MSI Principal Investigators Cheryl Gale (Associate Professor of Pediatrics and Genetics, Cell Biology, and Development), Dan Knights (Assistant Professor of Computer Science and Engineering), and Michael Sadowsky (Director, BioTechnology Institute; Professor of Soil, Water, and Climate) describe a new strategy to characterize intestinal fungal communities in infants. The authors used genomic techniques to study and characterize the fungal microbiomes of infants. They also obtained quantitative information about Candida, which is the dominant fungal genus in the human intestine. This work used software available through MSI as well as computational resources. The article can be read on the PLoS One website: Timothy Heisel, Heather Podgorski, Christopher M. Staley, Dan Knights, Michael J. Sadowsky, and Cheryl A. Gale. 2015. Complementary Amplicon-Based Genomic Approaches for the Study of Fungal Communities in Humans. PLoS One 10(2): e0116705. 10.1371/journal.pone.0116705.
Professors Gale, Knights, and Sadowsky use MSI resources to support their research into microbes and the microbiome. Professor Gale is using genomic approaches to establish how intestinal microbiome dynamics are associated with diseases in infants. Professor Knights is helping to create a bioinformatics pipeline for linking genetic variation with bacteria. Professor Sadowsky studies the metagenome of the Mississippi River, soil, and the human intestinal tract.
Image description: Relative Abundances of fungal taxa in infant fecal samples. Sample numbers (each from different infants) are noted on the lower line of the x-axis and the results of triplicate determinations are shown above the sample number for each. The most abundant taxa are indicated on the right-hand side of the graph, with low-abundance (present at <1.5% mean abundance across all samples) being grouped into the “Other” category. Sequencing reads are displayed as percentages of the total number of reads for each individual sequencing replicate. Image and description from Heisel, T, et al., 2015. PLoS One 10(2): e0116705. 10.1371/journal.pone.0116705.
posted on April 15, 2015
On Thursday, April 23, MSI held the sixth annual MSI Research Exhibition and dedicated our newest supercomputer, Mesabi. Mesabi is an HP distributed cluster featuring a large number of nodes with leading-edge Intel processors that are tightly integrated via a very high speed communication network. In addition, it contains a significant number of nodes with very large memory (up to 1 TB per node), accelerator nodes (GPU), and nodes with solid-state storage devices (SSD) for ultra high performance input/output.
At the Research Exhibition, MSI researchers presented posters of their work using MSI. The posters were judged by a panel of faculty members and prizes (travel awards) will be awarded. Posters competed in one of two categories, Physical Sciences and Engineering or Biological and Medical Sciences. Entrants were from a wide variety of disciplines. Ninety posters were submitted for the Research Exhibition, 81 of which participated in the competition. See photos from the event.
The judges awarded a Grand Prize winner in each category. The authors received a $1,500 travel award that can be used for conferences or professional meetings.
Biological and Medical Sciences Grand Prize: NINJA is not just another - short-reader mapper; authors: Gabriel Al-Ghalith, Emmanuel Montassier, Dan Knights; MSI PI: Dan Knights (Computer Science and Engineering)
Physical Sciences and Engineering Grand Prize: Equilibrium properties of DNA confined in nanochannels: A Monte Carlo chain growth approach; authors: Abhiram Muralidhar, Kevin D. Dorfman; MSI PI: Kevin D. Dorfman (Chemical Engineering and Materials Science)
Each category also had two Finalist prizes, who received $1,000 travel awards:
Biological and Medical Sciences Finalists:
Conformational dynamics of TNF receptors revealed by enhanced sampling techniques in molecular dynamics and free energy calculations; authors: Andrew K. Lewis, Anthony R. Braun, Christopher C. Valley, Jonathan N. Sachs; MSI PI: Jonathan N. Sachs (Biomedical Engineering)
Identification of novel candidate biomarkers and therapeutic targets in human sarcomas by systematic RNA sequencing; authors: Anne E. Sarver, Aaron L. Sarver, Venugopal Thayanithy, Subbaya Subramanian; MSI PI: Subbaya Subramanian (Surgery)
Physical Sciences and Engineering Finalists:
Combining memory- and CPU-intensive applications to examine geologic CO2 sequestration at micrometer to kilometer scales; authors: Benjamin M. Tutolo, Andrew J. Luhmann, Xiang-Zhao Kong, William E. Seyfried, Jr., Martin O. Saar; MSI PI: Martin O. Saar (Earth Sciences)
First-principles insights on line defect in strained NdTiO3 on SrTiO3; authors: Mehmet Topsakal, Jong Seok Jeong, Peng Xu, Bharat Jalan, K. Andre Mkhoyan, Renata M. Wentzcovitch; MSI PI: Renata M. Wentzcovitch (Chemical Engineering and Materials Science; MSI Fellow)
Research Exhibition attendees also voted on a People's Choice award. The poster receiving this award was Multiscale model of strain-dependent glomerular basement membrane remodeling; authors: Sarah Vanderheiden, Mohammad F. Hadi, Victor Barocas; MSI PI: Victor Barocas (Biomedical Engineering; MSI Fellow). These authors received MSI t-shirts.
Complete lists of the submitted posters in each category are linked below:
Image description: The images were taken from the Grand Prize-winning posters at the 2014 Research Exhibition.
Left: Inferred regulatory network of selected cardiac genes during differentiation. From “Inferring gene regulatory networks in cardiac differentiation by integrating temporal RNA-seq and ChIP-seq data.” Wuming Gong, lead author.
Right: A half-elipsoid mesh was used to model the isolated Pacinian corpuscle (PC). The ellipsoid had a major axis of length 1 mm and a minor axis of length 500 μ-m. Delaunay networks were rotated to become circumferentially aligned within the PC. 10 μm indentation of a cylindrical rod of diameter 250 μm occurred along the +z axis. From “Multiscale mechanical modeling of the Pacinian corpuscle.” Julia Quindlen, lead author.
posted on April 1, 2015
See all Research Spotlights.
Materials whose atoms form a crystal structure - the atoms are organized in a symmetric spatial arrangement - are useful for various applications, such as electronics. During the crystal growth process, dislocations can form. One type is called a “screw dislocation,” which can be seen in the image above. Defects in a crystal will affect its properties and thus have implications for the material’s use.
MSI Principal Investigator Traian Dumitrica, an associate professor in the Department of Mechanical Engineering (College of Science and Engineering), post-doctoral fellow Dr. Yuxiang Ni, and their colleagues at the Ecole Centrale Paris have been studying screw dislocations by performing equilibrium molecular dynamics simulations on MSI’s supercomputers. In a recent paper, these researchers discussed their finding that the area around a screw dislocation in a silicon carbide crystal has higher internal thermal resistance than the rest of the material. This is interesting theoretically, and the findings also can be applied to the design of materials useful for high-temperature electronics and thermoelectric applications. The paper was featured on the cover of the prestigious journal Physical Review Letters in September 2014. (Y Ni, S. Xiong, S. Volz, T. Dumitrica. Thermal transport along the dislocation line in silicon carbide. 2014. Physical Review Letters. 113, 124301).
Professor Dumitrica and his research group use MSI resources for molecular dynamics simulations related to studies of the mechanical properties, stability, and behavior of nanoscale materials. A previous paper was featured in a Research Spotlight, “Modeling Properties of Graphene,” in August 2014.
Image description: Supercells for the SiC calculations: pristine (left), 1b (center) and 2b (right) screw dislocated (b = 3.08 Å). The heat carrying direction is z. The cross-section dimensions are 4 x 4 nm. Length is 7 nm. Image and description Ni, Y et al., PRL, 2014, 113:124301. ©2014 American Physical Society
posted on March 18, 2015
In the fall of 2014, MSI added a Ceph object storage system as an option for second-tier storage for MSI users. Many MSI researchers are in disciplines that use huge amounts of data, such as informatics, genomics, and astrophysics. Often, much of this data does not need to be stored on a machine with high-speed access. This is where a good second-tier storage option is valuable.
The Ceph system currently has 1.4 PB of space for data storage. Its access features allow researchers to exchange data with colleagues outside the University of Minnesota, port cloud-based workflows to MSI systems, and store inactive data separately from the high-performance systems. Several MSI researchers, shown below, have already begun using the Ceph system.
Professor Shaul Hanany (Physics and Astronomy/Minnesota Institute for Astrophysics; College of Science and Engineering) and his research group are analyzing data from the E and B EXperiment (EBEX), a NASA-funded balloon-borne polarimeter designed to measure the polarization of the cosmic microwave background and to develop methods for subtraction of foreground sources from these data. The first science flight of EBEX collected 1 TB of raw data. The Hanany group is performing simulations to understand the uncertainties and systematic effects associated with the data analysis.
Assistant Professor Suzanne McGaugh (Ecology, Evolution, and Behavior; College of Biological Sciences) uses MSI to support genomics and transcriptomics studies in cavefish, reptiles, and other animals. A recent Research Spotlight featured a paper by Professor McGaugh and her colleagues that disclosed the first de novo genome assembly for the cavefish Astyanax mexicanus, the Mexican tetra fish.
Assistant Professor Peter Morrell (Agronomy and Plant Genetics; College of Food, Agricultural, and Natural Resource Sciences) is studying the effect of domestication and strong selection during crop improvement on the level of deleterious mutations in self-fertilizing crops. This involves DNA and RNA sequence assembly, identification of single nucleotide polymorphisms, and using probabilistic approaches to determine the proportion of mutations that are likely to be deleterious.
All these groups have placed terabytes of their data onto the Ceph system. This has freed up space in each group’s allotted quota on the high-performance storage systems, creating a more efficient use of those systems.
posted on March 4, 2015