High-Performance Computing

Zorro, the AU High-Performance Computing (HPC) System supports the efforts of faculty and student researchers working with computationally intensive projects. The AU HPC System provides high-speed parallel/distributed data processing and research-dedicated storage with high input-output capabilities. Hardware includes:

  • blade chassis (Dell PowerEdge M1000E)
  • 16 Dell PowerEdge M630 blade servers each housing 2 Intel Xeon
  • CPU E5-2650 v4 @ 2.20GHz processors (12 cores per socket), 128
  • GB of memory, and a 800 GB SSD
  • Dell PowerEdge R730 server with Nvidia Tesla K80 GPU, 128 GB of RAM. 800 GB SSD
  • Infiniband interconnect
  • 90 TB Isilon Network Attached Storage cluster

This structure allows researchers to significantly reduce computational time by breaking down large jobs into smaller, independent units, and simultaneously executing them. Jobs that usually take days or weeks to run can now be completed in a matter of a few hours.

HPC Image, Blue lines

Research Support Group Consulting Services

Our staff can provide educational and consulting support. General assistance is available for users of the HPC cluster as well as help with technical problems, such as code optimization. We offer group tutorials for HPC users and individual one-on-one consulting support at all phases of research. Training is available at CTRL or at your location.

In order to find out more about the cluster and obtain a user account, please go to http://www.american.edu/cas/hpc/.

For any queries, questions, and comments on HPC, please contact Angel Bogushev or call x2797.

HPC Software

Licensed: Matlab Distributed, Mathematica, STATA MP

Open Source: C++, CUDA, Enthought Python Distrubuted, Fortran, MEGA, Mr. Bayes, Muscle, Perl, Octave, OpenBUGS, R-Project

HPC Seminar Series

Homology Modeling of G Protein-Coupled Receptors

Professor Stefano Costanzi, Department of Chemistry (CAS)

Thursday, March 9th, 2017 | 12:00 p.m. | Hurst Hall, Room 203

G protein-coupled receptors (GPCRs) are a family of about 1,000 membrane proteins, which are targeted by a large share of the marketed drugs. In light of their pharmaceutical relevance, there is substantial interest in shedding light onto their three-dimensional structure through X-ray crystallography to assist the drug discovery process. However, to date, exper­imental structures have been solved only for about 30 members of the GPCR family. In our work, we employ a technique known as homology modeling to build three-dimensional models of GPCRs with unknown structures using receptors with solved structure as templates. With this seminar, I will illustrate the result of a quantitative study through which we proved the modellability of GPCRs. Moreover, I will discuss how the HPC system can be exploited for their construction of homology models.

The Subterranean Genome of the Devil Worm

Professor John R. Bracht, Department of Biology (CAS)

Friday, October 7th, 2016 | 12:00 p.m. | Hurst Hall, Room 203

The subterranean worm H. mephisto was first discovered in a gold mine in South Africa, living nearly a mile underground in water-filled cracks in the earth’s crust. Completely isolated from the terrestrial biosphere, this organism has managed to survive, and thrive, under conditions that had been considered lethal to complex life. In this talk I will present recent data from whole-genome sequencing and analysis, and discuss how this finding sheds light on adaptive change in evolution, the limits of complex life on earth, and even on the search for life on other planets. I will discuss the role of Zorro in this project as well as my use of high-performance computing in genomics instruction.

This talk will be geared towards a non-specialist audience.