Utilization of HPC in Cancer Research Dave Billiter

Utilization of HPC in Cancer Research Dave Billiter, PMP Director-Research Informatics Core The Research Institute at Nationwide Children’s Hospital, Columbus, OH

Presentation Overview l Introduction l Research Informatics and Imaging Core l Infrastructure l Image Analysis l Hypothesis l Microsoft HPC l Conclusion

The Research Institute at Nationwide Children’s Hospital

Research Informatics Core Mission The Research Informatics Core of The Research Institute at Nationwide Children’s Hospital, under the vision of Dr. Stephen J. Qualman and direction of Dr. Nilsa Ramirez, is an informatics application development team dedicated to increasing cure rates and decreasing side effects in therapy through innovation of information technology. http: //informatics. nchresearch. org/

Informatics Core Team l l l Dave Billiter, PMP- Informatics Director Tom Barr- Imaging Manager Bob Neville- Senior Software Developer Scott Hargrose- Software Developer/DBA Bill Beyer- Software Developer/Business Analyst Heather Bao- Software Developer Heather Day- Project Manager Subha Krishnamachary- Software Developer Andy Fox- Software Developer Mark Plaskow- Software Architect Eric Kramer- Data Architect Jason Weis- QA/QC

Informatics Core Capabilities l Custom Application Development l Informatics Consulting l Solution Analysis l Data and Environment Integration l Project Management l Systems Architecture l Web Site Development

Informatics Core Technology l l l Microsoft. NET Framework (. NET Suite) MS SQL Server 2000/2005/2008 (6. 5/7) MS Access 97/2000/2003/2007 MS Team Foundation Server My. SQL Oracle Cold. Fusion Java Ruby/Ruby On Rails Aperio Scan. Scope and Image Software Linux/Apache/PHP/JSP Freezerworks

Informatics Core “Clients” l l l Center for Childhood Cancer Investigators Children’s Oncology Group (COG) Gynecologic Oncology Group (GOG) Cooperative Human Tissue Network (CHTN) Childhood Cancer Survival Studies (CCSS) Cancer Therapy Evaluation Program (CTEP) l l l l Center for Innovation in Pediatric Practice Biobehavioral Health Nationwide Children’s Hospital NCI Group Banking Committee (GBC) NCI Cancer Genome Atlas Project Biopathology Center (BPC) The Lance Armstrong Foundation

Imaging Core Multidisciplinary team Imaging services to enhance biomedical research for multiple national cooperative groups and other researchers l Currently work with over 30 pathologists l Focus Areas l Pathology Imaging l Automated Pathology Review Applications l Digital Slide Conferencing l Image Architecture Analysis & Development l Digitized Pathology Reports l Online capture of pathology review data l l

Infrastructure-Hardware Ohio Supercomputer Center (OSC) l IBM x 3455 1 U server featuring 1 dual-core AMD 2. 4 GHz Opteron processor, with 2 GB of main memory, and a pair of 70 GB mirrored SATA drives for system software and swap space. l Motherboard has a total of 4 processor sockets, and is capable of being populated with quad-core processors in each socket, for a total of 16 cores. l Linked to a DDN 8500 SAN (25 TBs), provides storage for application executables and data.

Infrastructure - Software The operating system is Microsoft HPC Server 2008, based on Windows Server 2008 64 -bit Edition l Built to scale to thousands of cores l Features cluster management tools, support for third-party clustered file systems, advanced failover capabilities, service-oriented architecture (SOA) job scheduler. l

Infrastructure - Software Also includes remote direct memory access interface, and cluster interoperability through standards such as High Performance Computing Basic Profile specification produced by the Open Grid Forum. l Combines the operating system with a message passing interface and a job scheduler into a single package. l

HPC Cancer Research-Imaging Image Analysis l Anaplasia l Mitotic Index l Necrosis l Impact of Chemotherapy in Radiology Modalities

Whole-Slide Image l Tissue Microarray l Several patient’s materials on one glass slide l Easily compare staining characteristics of several tumors

Image Analysis l Whole-Slide (Image) Search

Image Analysis l Whole-Slide Image Processing

Image Analysis l Pattern of pixels for Nuclei and ability to locate abnormal clustering.

Image Analysis l Algorithms applied to both solid and liquid diseases.

HPC Cancer Research Hypothesis l The association and utilization of a high performance computing environment within cancer research will enhance detection and optimize the pathology review process thus allowing for rapid and specific patient therapy.

Why Microsoft HPC? INTEGRATION l Current Windows based vendor applications. l Familiarity with Informatics development environment. l Windows access to image files stored in Linux-based storage share. l Performance.

Microsoft HPC Challenges Local Microsoft HPC subject matter expertise. l DEP (Data Execution Prevention) l MS OS’s like Vista and Server 2008 are very (overly? ) security conscious and install with all security features enforced – including DEP! l Error messages from application services that refused to start/stay running did not relate to DEP; implication of DEP had to be deduced.

Accomplished l Successful installation and execution of Aperio apps (DSC, Spectrum, Image. Scope, Image. Server). l Successful installation and execution of Apache web server, configuration of Apache to co-exist with IIS. l Configuring path to image storage shares.

In Process and Future Activities Installation of Aperio Data. Server, configuration of Data. Server and Image. Server. l Will need to modify algorithms to write analysis results to Aperio database & relate to other image metadata. l Modify Aperio database to accept additional algorithm output. l Modify Spectrum to display analysis metadata, easily reference annotated images. l

Acknowledgements l Brian Hammond- Microsoft HPC l Kevin Wohlever- OSC l Lin Li- OSC l Bill Beyer- RIC l Mark Plaskow- RIC l Tom Barr- Biopathology Imaging Core l Dr. Kathy Nicol- Natiowide Children’s Hospital

Conclusion l Questions & Answers THANK YOU!!!!