NIH | National Cancer Institute | NCI Wiki  

Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

Reliable, innovative solutions that make it easy for NCI investigators to use high performance computing.  

 

 

Round Rectangle
bgcolor#33B8FF#E6F8FF
titleHPC Program Overview
Horizontal Navigation Bar
titleSite Map
url - Consultation and support about data-oriented resources that are available for NCI researchers to move, share, and store data.

 

Horizontal Navigation Bar Page
titleServices

Services


Button Hyperlink
iconsearch
titleHPC Consultation
typestandard
urlHPC Consultation
- Consultation, training, and support for high performance computing systems used by NCI researchers.

 

Button Hyperlink
iconsearch
titleHPC Application Development
typestandard
urlHPC Application Development
- High performance (HPC) and scientific computing application development for NCI researchers.


Button Hyperlink
iconsearch
titleHPC Data Transfer Resources
typestandard
urlHPC Data Transfer Resources
- High performance and scientific computing data transfer resources for NCI cancer research studies/projects.
Button Hyperlink
iconsearch
titleHPC Research Computing and Data Analysis
typestandard
HPC Research Computing and Data Analysis
Horizontal Navigation Bar Page
titleBackground

Background

 

Biomedical research computation enables a growing number of scientists and clinicians to analyze, use, manage, and share their data—and their discoveries.

At NCI, wehave developed innovative High Performance (HPC) and Scientific Computing services that help investigators accelerate and advance their cancer research.

 

Our services include HPC support and consultation, education, and reliable access to various state-of-the-art HPC systems including the NIH Biowulf system, the NCI MOAB system, and HPC programs at the Department of Energy.  

 

We also work closely with investigators across NCI to develop, optimize and/or validate HPC applications—and computing technologies—that enable rapid data transfer and provide effective data management. With ongoing input from the cancer researchers we serve, we continue to improve and evolve our HPC services, capabilities and opportunities. As partners to cancer researchers worldwide, we are guided by investigators’ challenges and opportunities.

 

Above all, we seek to expand scientists’ use of high performance computing to accelerate discoveries in predictive oncology research and advance effective methods of prevention, treatment, and cure. 

 

Expand
titleContext

CBIIT HPC Strategy

With a focus on providing robust and reliable solutions enabling NCI investigators to utilize HPC in their efforts, the CBIIT HPC strategy is focused on these interconnected areas:

  • Working closely with investigators across NCI to enable broader utilization of HPC through HPC training, education and reliable system access
  • Guided by investigator challenges and opportunities, provide support and consulting for HPC needs as well as development, optimization and/or validation of HPC applications useful to cancer research and clinical applications
  • Effective data management and information delivery solutions in support of HPC applications used in cancer research
  • Exploration and evaluation of emerging HPC technologies for use in cancer research, information delivery and data center operations
  • Developing essential partnerships within NCI, NIH, HHS, government, academically, commercially, nationally and internationally fostering the expanded use of HPC in cancer research
  • Develop and deliver services supporting current and future HPC needs of NCI investigators
  • Continual incorporation of investigator input to improve and evolve HPC services, capabilities and opportunities

 

Long Range Guiding Objectives for HPC in Cancer Research

With guidance and insight provided by the cancer research and clinical community within NCI, deliver robust, reliable HPC capabilities and support that:

  • Enable broader understanding of cancer, cancer system dynamics and cancer characterizations
  • Enable rapid identification of potential cancer risks and presence of cancer in individuals
  • Enable rapid determination of optimal treatment options for patients
  • Expand treatment options through improved discovery and rapid, reliable validation
  • Foster computational integration and cooperation across the global cancer research community
  • Enable transfer and flow of HPC technologies between NCI and other stakeholders
  • Enable NCI to take full advantage of computational advances to accelerate cancer research

 

Foundations for Successful HPC in Cancer Research and Clinical Development

  • Useful
    • Performing needed functions and delivering key capabilities
    • Enabling technologies can be eventually used in clinical application
  • Reliable
    • Assuring new computing technologies and applications are functionally reliable
    • Assuring technologies and applications are validated and verified
    • Assuring appropriate reproducibility of delivered solutions over time
  • Adaptive
    • Exploring new and emerging technologies and applications for use in cancer research and translation
    • Utilizing multiple sources of input (internal and external) to improve overall HPC capabilities
  • Portable
    • Enabling intellectual investments to transition across emerging and evolving technology platforms
  • Efficient
    • Providing solutions in as rapid as possible yet in a cost-effective manner
Horizontal Navigation Bar Page
titleCommunications

Communications


Button Hyperlink
iconsearch
titleHPC Thought Leaders Presentations
typestandard
urlHPC Thought Leaders Presentations

 

HPC Presentation to NCI Informatics & IT Advisory Group - Eric Stahlberg - 10.20.16

Horizontal Navigation Bar Page
titleTraining and Education

Training and Education


HPC in Biomedical and Life Science Applications

Expand
  • High Performance Computing (HPC) has an invaluable impact on driving advances in cancer research. Projects conducted within the context of the Precision Medicine Initiative, the National Cancer Institute, the Department of Energy pilots, and the National Strategic Computing Initiative are only examples of how new frontiers can be pushed using the next generation of supercomputers.
  • Areas of impact include, but are not limited to:
    • predicative algorithms for cancer therapy using machine learning, 
    • molecular dynamics simulation modeling for drug discovery, and 
    • medical image analysis.
  • This presentation highlights different programming models and run times that expose parallelism on multicore/multiprocessors platforms such as shared and distributed memory architectures, and Graphics Processing Units (GPUs). Target audience are computational and data scientist looking to do more with available HPC resources and prepare for capabilities of even larger systems than those currently available.
  • An overview on the projects conducted at the NCI is given. This includes the fields of molecular dynamics simulation, analysis of transcript slicing, and image segmentation and analysis.  Some best practices and lessons learned will also be shared.

Button Hyperlink
iconsearch
titleHelpful Links
typestandard
urlHPC Related Links

 

Questions?

General Support : Miles Kimbrough | miles.kimbrough@nih.gov | 240.276.5251

Consultation and Guidance : Eric Stahlberg | eric.stahlberg@nih.gov | 240.276.6729

Technical Support : George Zaki | george.zaki@nih.gov | 240.276.5171

 

 

 

 

 

 

 

 

Round Rectangle
bgcolor#90F999#E3FFE6
titleMonthly Updates
Expand
titleClick Here for Recent Updates

Table of Contents

Table of Contents
maxLevel4
minLevel3

3/6/17

  • High Performance Computing User Group & Office Hours – The HPC Program is developing a user group with the objective of serving those requiring HPC assistance, and expanding the HPC community within the NCI intramural community.  The program will be implementing office hours in Frederick on a recurring basis, to serve as points of interaction among the developing HPC community and to aid those with HPC-specific needs or questions.
  • CANDLE Workshop – Serving as one of the key computational frameworks to support the NCI DOE Collaboration, the CANcer Distributed Learning Environment is designed to use machine-learning algorithms to find patterns in large datasets with the goal of offering insights that may ultimately result in improved cancer treatment.  Using this computational architecture, participating DOE labs are focused on accelerating methods to identify promising new treatments; deepening understanding of cancer biology; and understanding the impact of new diagnostics, treatments and patient factors in cancer outcomes.  
    • There will be a workshop on April 18-19 at the NIH which will provide an opportunity to learn more about deep learning, share insights into research, and explore the potential through open-ended discussions and hands-on activities.
    • Stay tuned for more information and registration details – please contact Miles Kimbrough (miles.kimbrough@nih.gov) or Eric Stahlberg (eric.stahlberg@nih.gov) with questions or to discuss further.
  • Globus user-focused Webinar – Globus, a cloud-authenticated data management and transfer platform, will be hosting a user-focused webinar in mid-April to benefit those interested in exchanging datasets across a variety of sources.  The webinar will provide a high-level overview of Globus, steps to start using the service, and common use cases along the following topics:
    • When, where, and why to use Globus?
    • NIH account specs - distinction from Globus Plus
    • What do system administrators need to set up managed endpoints?
    • Which endpoints are already set up?
    • How to set up Globus on your own desktop
    • How to transfer and share data
    • If sharing with collaborator, what info does collaborator need?  What do you need to give to collaborator?
    • New Globus command line interface, allowing users to script their transfers
    • Encryption, verification, and expected data transfer speeds as compared to other resources (e.g. FTP)
  • More details are forthcoming – in the meantime please contact Miles Kimbrough (miles.kimbrough@nih.gov) to discuss further.


2/6/17

  • HPC Data Management Environment 1.0.0 Release – The HPC data management environment (DME) provides a number of application programming interfaces (APIs) to support common scientific data/metadata management functions across the NCI.  Currently in initial release,  HPC DME archive storage can be a permanent storage solution for user data, and can be used as a platform to search, manage and transfer data onto other storage systems.  Common use scenarios may be characterized as follows:
    • Register a collection (PI Lab, Project, Run, Sample or dataset)
    • Register a single data file/object into storage archive synchronously/asynchronously
    • Perform an update on a metadata attribute
    • Subscribe to a known event
    • Generate a report
    • Update/assign permission
    • Perform simple search functions
    • Download a data file/object to Globus share or local directory
  • To obtain an HPC DME account, please contact the DME team at HPC_DME_Admin@nih.gov.  Additional details on the release can be found at https://wiki.nci.nih.gov/display/HPC/HPCS-0006
  • For any other questions, please contact either Miles Kimbrough (miles.kimbrough@nih.gov) or Eric Stahlberg (eric.stahlberg@nih.gov)
  • 2017 Globus Hackathon – Globus, a cloud-based data management and transfer platform, hosted a Hackathon in January at the NIH main campus to benefit those interested in exchanging datasets across a variety of sources.  The event provided hands-on training, sample code writing, and breakout sessions to the more than 40 attendees.  As the event was very well-received, and due to increasing demand, scheduling efforts are now underway for an NIH-wide user-focused webinar to provide a high-level overview of Globus and steps to start using the service.  More details are forthcoming – in the meantime please contact Miles Kimbrough (miles.kimbrough@nih.gov) to discuss further.


1/10/17

  • 2017 Globus Hackathon – Globus, a cloud-based data management and transfer platform, will be hosting a Hackathon at the NIH on January 24-25.  This event will benefit those interested in exchanging datasets across a variety of sources and will be separated into two training sessions based on prior technical experience – a high-level overview for researchers and system administrators, and a developer-focused session.   The Hackathon is free to NIH employees and will include hands-on training, sample code writing, and breakout sessions to promote cross-collaboration.  Register here and review the agenda here.
  • CBIIT TechScouts – CBIIT TechScouts currently serves the CBIIT community as an idea-sharing platform which allows members to share new tools, technologies, and methodologies relevant to added workplace efficiency.  In order to promote broader community engagement and increase awareness of this platform, the TechScouts will be undergoing some exciting new developments throughout 2017.  More updates will follow soon – in the meantime, please contact Miles Kimbrough (miles.kimbrough@nih.gov) to discuss further.


12/16/16

  • SuperComputing 2016 - As a follow-up to the 2016 International Conference for High Performance Computing, a blog has been made available on the NCIP Hub which summarizes NCI's involvement in the conference and sheds light on the challenges and opportunities that lay ahead.  The blog can be found at https://ncip.nci.nih.gov/blog/precision-medicine-inspires-hpc/.

12/8/16

  • Frontiers of Predictive Oncology and Computing Meeting (FPOC)– With over 100 attendees from across the Department of Energy, the National Cancer Institute, academia, industry and other government agencies, the FPOC meeting (hosted by Intel July 12-14, 2016) provided an opportunity to gain insight into challenges and opportunities for the future. The white paper summarizing the meeting is now available through Intel’s website and can be found at http://www.intel.com/content/www/us/en/government/predictive-oncology-and-computing.html.
  • SuperComputing 2016 – The 2016 International Conference for High Performance Computing welcomed NCI CIO, Dr. Warren Kibbe and other representatives from CBIIT in Salt Lake City, UT, November 12-18, for a series of presentations and workshop sessions.  These included a Computational Approaches for Cancer workshop, a Plenary Session with Dr. Kibbe serving as a panelist, and a Birds-of-a-Feather session.  Electronic copies of these presentations will be circulated over the next month – more information on the conference can be found at http://sc16.supercomputing.org/.
  • 2017 Globus Hackathon – Globus, a cloud-based data management and transfer platform, will be hosting a Hackathon at the NIH on January 24-25.  The Hackathon is free to NIH employees and will include hands-on training, sample code writing, and breakout sessions to promote cross-collaboration.  Register here and review the agenda here.
  • Communications about HPC in NCI – To reflect current and emerging efforts of the NCI’s collaboration with the DOE, a landing page has been created at http://hpc.cancer.gov which will provide an anchor point for new developments and resources related to the collaboration.
  • HPC Education and Training – As a part of ongoing outreach efforts, The HPC Program will be coordinating a PI-level presentation with the CCR in early January.  Primary presentation will include application domains, NCI DOE collaborative updates, and Big Data management, among others.  For more information, please contact Eric Stahlberg (eric.stahlberg@nih.gov) or George Zaki (george.zaki@nih.gov). 

10/3/16

  • High Performance Computing (HPC) Education & Training – The HPC Program will be sponsoring a Graphics Processing Unit (GPU) Training Workshop on Tuesday, October 25th, in collaboration with CIT.  The workshop will cover the fundamentals of GPU architecture and programming through hands-on exercises.  Please contact either George Zaki (George.zaki@nih.gov) or Miles Kimbrough (miles.kimbrough@nih.gov) for more information.
  • 2017 Globus Hackathon – Globus, a cloud-based data management and exchange platform, will be hosting a Hackathon at the NCI in late January.  The Hackathon will include hands-on training, sample code writing, and breakout sessions to promote cross-collaboration.  Stay tuned for further information and registration details.
  • HPC and Data Management Support – For assistance with Globus, GPUs, and other high-performance computing or data management needs, visit NCI at Your Service (service.cancer.gov) and simply submit a request for support using the ‘High Performance Computing / Data Management’ request button. Or, simply email George Zaki (george.zaki@nih.gov) or Miles Kimbrough (miles.kimbrough@nih.gov) to get started. 

 

9/6/16

  • Archive Data Management – A new archive data service for large data is now available for evaluation. The new service provides application interfaces to readily access backend storage technologies such as the new Cleversafe storage moved into production last month. Individuals and groups interested in learning more about how this resource may benefit current initiatives may reach out and contact Eric StahlbergMiles Kimbrough or George Zaki
  • Education and Training -  Learn more about how high-performance computing (HPC) can be used to accelerate cancer research and clinical applications. Individuals and groups interested in learning more about HPC, either in general or with specific technologies and scientific challenges in mind may reach out and contact Eric Stahlberg, Miles Kimbrough or George Zaki.
  • The upcoming Computational Approaches for Cancer workshop scheduled for November 13, 2016 as part of the International Conference for High Performance Computing, Networking, Storage and Analysis has extended a call for extended abstracts until September 15, 2016. More information can be obtained at the link http://www.scworkshops.net/cancer2016/.

8/8/16

  • Frontiers of Predictive Oncology and Computing Meeting - With over 100 attendees from across the Department of Energy, NCI, academia, industry and other government agencies, the meeting (hosted by Intel July 12-14, 2016) provided an opportunity to gain insight into challenges and opportunities for the future. A white paper summarizing the meeting is to be developed.
  • New Data Services with Cleversafe – The Cleversafe storage system officially was moved into a production operational status at the beginning of August. Led by the IT Operations Group at Frederick National Laboratory and working with many stakeholders including CCR, CBIIT and NIH CIT, the new system is used within industry and in key efforts such as the Genomic Data Commons to provide a high level of data assurance for archive and stable data. Stay tuned for further information on opportunities to learn more how this new resource may benefit your scientific and operational needs.
  • Education and Training - Plans are underway to develop educational opportunities to learn more about how high-performance computing (HPC) can be used to accelerate cancer research and clinical applications. Individuals and groups interested in learning more about HPC, either in general or with specific technologies and scientific challenges in mind may reach out and contact Eric StahlbergMiles Kimbrough or George Zaki.
  • Computational Approaches for Cancer workshop - Scheduled for November 13, 2016 as part of the International Conference for High Performance Computing, Networking, Storage and Analysis. A call for papers has been issued. More information can be obtained at the link http://www.scworkshops.net/cancer2016/
Round Rectangle
bgcolor#F9DC90#FFF8E3
titleBlog Posts
Expand
titleClick Here for Blog Posts

Blog Posts
reversetrue
contenttitles