Malaysia Genome Institute advances bioinformatics research using SGI Technology

Feb 21 2008

To create an optimal workflow that encompasses the diverse computing requirements of bioinformatics research, the Malaysia Genomics Institute (MGI) selected SGI to design a unique SGI hybrid computing solution.

Because bioinformatics research uses a variety of software at various stages in the workflow, the combined use of shared-memory systems and clusters can provide an optimal environment for the efficient use of each application. MGI looked to SGI to design a seamless, cohesive workflow for the entire process that would integrate into the existing infrastructure. SGI delivered a hybrid computing solution that consists of the new SGI BioCluster as well as an SGI(R) Altix(R) shared memory server and an SGI(R) InfiniteStorage CXFS(TM) SAN shared file system, which makes the entire workflow transparent to the user regardless of software employed or stage of the process at which scientists are performing.

"Bioinformatics brings together an avalanche of systematic biological data with the analytic theory and practical tools of mathematics and computer science," said Mohd Noor Mat Isa, Laboratory Manager, Malaysia Genome Institute. "SGI delivered flexible computing with a hybrid architecture that integrates high-throughput, high-performance and FPGA-based solutions for all of our bioinformatics applications -- a solution that seamlessly runs applications in the environment best suited for optimal performance."

Genomic sequencing operations utilize applications such as PHRAP, Staden Package, JAZZ , Consed, and BLAST each with specific compute requirements necessitating multiple systems. The SGI BioCluster solution includes a software layer providing transparent workload scheduling through which users can submit jobs without hardware concerns. The SGI solution's administrative software intelligently cues submissions to run on the appropriate platform, providing best-in-class performance and optimal load balancing. Integrated cluster management tools and dynamic software controls can change the total system from a single meta-cluster to semi-autonomous sub-clusters as needs dictate. For unexpected time-critical computing challenges, the SGI hybrid solution enables administrators to bring all available resources to bear on a single job.

MGI, also known as GENOMalaysia, is a network-based organization carrying out discovery research on tropical bioresources through projects in genome sequencing, comparative and functional genomics, and structural biology. Several major research projects are currently underway at MGI. For example, the Microbial Genomics Research for Gene and Natural Product Discovery project is focusing on two model organisms, Burkholderia pseudomallei, a locally important soil pathogen and Eimeria tenella, a local avian protozoan. The project will harness the SGI solution to dissect the organisms' genomes to gain new insights on virulence, antibiotic resistance, host specificity and infectivity.

The Institute also collaborates with other research facilities and universities throughout the world, including the Antarctica Research Group at the University of Tasmania, Wellcome Trust Sanger Institute, University of Sheffield, UK and Stanford University.

SGI worked with channel partner Quantum Beez Sdn Bhd (formerly known as Open Source Systems) to design and develop the SGI hybrid solution for MGI.

"Malaysia Genome Institute needed a single, integrated computing platform that leverages the best of two distinct computing environments: a larger shared memory system for high performance computing, and a cluster-based system for high throughput computing, both supported by an SGI InfiniteStorage CXFS SAN shared file system that provides uncompromised data sharing, improved workflow, and reduced costs for these data intensive environments," said Deepak Thakkar, higher education and research segment manager, SGI. "SGI produced one powerful solution with two architectures and a high-speed SAN infrastructure for direct access to all data. This hybrid solution delivers optimum return on investment for heterogeneous or multi-vendor based workflows."