High Performance Computing (HPC) at Dell
Thanks for visiting our online HPC and technical computing community. We have an active group of contributors from our technical Dell engineers, to our industry leading customers, and worldwide partners.
Stay awhile, browse some of our great content, and join the conversation!
Introduction by Dell's Elliott BergerOur team develops High Performance Computing solutions for our customers. Considering Dell's origins as an expert low cost, build-to-order, PC provider, many people think of Dell for low cost, commodity servers as a platform for building an HPC cluster. To their surprise, the truth is far greater as we are intensely focused on providing comprehensive HPC solutions--hardware, software, services to shorten the time to discovery, simplify management and use, and squeeze costs to increase the value of HPC systems to all users who depend on them to advance their cause.
One of the interesting requirements that increasingly falls in our scope is the use of visualization. The HPC environment at the University of Louisville is a general resource, used by a variety of departments to further research and discovery. In recognition of the insatiable need for compute cycles to address the complex challenges under study there, they smartly built the environment with scalability and growth in mind. In addition to the compute cluster, they also built a visualization wall. Another general resource on campus, this "viz wall" can be utilized to enable researchers to literally see relationships that may be hard to spot in numeric data.Below, Dr. Rouchka describes his work at the intersection of computer science and biology along with that of his colleagues at the University of Louisville. The advances they are making towards personalized medicine are promising and exciting. Their use of the viz wall to view and study results generated by the HPC cluster is very interesting and representative of the growing scope of the extent of our challenge on the Dell HPC team. -- Elliott Berger
The Archon X Prize for Genomics was developed in 2006 as a $10,000,000 award to the first team “that can build a device and use it to sequence 100 human genomes within 10 days or less with an accuracy of no more than 1 error in 100,000 base pairs, with sequences accurately covering at least 98% of the genome, and at a demonstrated cost of no more than $10,000 per genome.” (http://genomics.xprize.org). Advances in sequencing technologies such as those developed companies such as Roche, Illumina, Life Technologies, Applied Biosystems, and Pacific Biosystems have made the personalized aspect of P4 MedicineTM (predictive, preventive, personalized, and participatory) a very real possibility in the not-to-distant future. Current projects such as the Personal Genome Project (http://www.personalgenomes.org/) , the International HapMap Project (http://hapmap.ncbi.nlm.nih.gov/), and the Pediatric Cancer Genome Project (http://www.pediatriccancergenomeproject.org) are currently underway to sequence multiple partial and/or complete genomes. Each of these projects brings great data storage needs. In the case of the Pediatric Cancer Genome Project which proposes to sequence 600 healthy and 600 cancer genomes, the storage space for the sequence data alone is estimated to consume several petabytes (http://www.stjude.org/SJFile/GenomeProm.pdf). As a result, a paradigm shift is beginning to be observed in bioinformatics where the location of the data remains static, and the location of the algorithms used to analyze sequence data is dynamically moving to the location of the data.Each of these sequencing projects brings with it a great potential for unlocking knowledge concerning how genetic differences play a role in disease. At the University of Louisville, my lab, in collaboration with Dr. Ted Kalbfleisch of the Department of Biochemistry and Molecular Biology, is looking specifically at HapMap trios which consist of a complete genome of an individual along with compete genomes of each of their parents, in order to detect novel retrotransposon events. We are also searching for events that may not be de-novo within the trio, but have been previously uncharacterized within the human reference genome. We are working to use our 18 display, 6x3 Dell visualization wall (incorporating Dell PowerEdge 2950 servers and Dell Ultrasharp 3007WFP wide screen monitors) to visualize the annotations and put them in a larger context. In addition, our lab is collaborating with Dr. Nigel Cooper of the Department of Anatomical Sciences and Neurobiology in order to discover and annotate transcript variants, particularly transmitter related receptors which could eventually be of considerable interest to the pharmaceutical industry. We are also working together to understand differential gene expression patterns of these isoforms within tissues associated with the central nervous system. It is important to consider both how much a gene’s expression changes along with the localization of the expression. Thus, Dr. Cooper’s lab has provided high-resolution in-situ hybridization images which contain individually labeled gene isoforms. We are working on a system, Ymage, which will allow researchers worldwide to share repositories of images in a federated manner and will provide mechanisms to annotate the images in a distributed fashion. We are developing an application for Ymage specifically for the visualization wall that will allow researchers to see the whole context of their high resolution images (up to 15,360 by 4,800 pixels) along with user annotations and image registration to available anatomical atlases. The age of personalized genetics provides great opportunity for bioinformaticians as well as computational and data storage challenges. We look forward to all of the possibilities that lie ahead in the near future.What are your thoughts? We'd love to hear your comments.View related video here: