Skip Over Navigation Links
National Institutes of Health
:

Interoperability Workshop

JANUARY 8, 1999
MARRIOTT HOTEL, POOKS HILL
Kensington Room
Bethesda, MD
8:30am – 3:30 pm

Agenda

8:30 Meeting Introduction-All

9:00 Overview of Issue and lessons learned from large database operations

(National Center for Biotechnology Information, National Library of Medicine, NIH)

10:00 Break

10:15 General Discussion-All.

12:00 Lunch Break

1:00 Summary Discussion of Interoperability and Future Steps-All

2:00 Other Issues-All

3:30 Meeting Adjourn

Participants Neuroinformatics
Interoperability Workshop

Consultants

Ira Black- UMDNJ-Robert Wood Johnson Medical School

Joseph T. Coyle-Harvard Medical School

Daniel Gardner-Cornell Medical College

Edward D. Jones-University of California, Davis

Perry Molinoff-Bristol-Meyers Squibb

Gordon Shepherd-Yale Medical School

NIH Participants

Tom Aigner-National Institute on Drug Abuse

Connie Atwell- National Institute of Neurological Disorders and Stroke

Dennis Benson- National Center for Bioinformatics, National Library of Medicine

Gerald Fischbach-Director, National Institute of Neurological Disorders and Stroke

Michael Hirsch-National Institute of Mental Health.

Steven Hyman-Director, National Institute of Mental Health

Cheryl Kitt- National Institute of Neurological Disorders and Stroke

Stephen Koslow-National Institute of Mental Health. Chair of the Federal Interagency Coordinating Committee on the Human Brain Project(FICC-HBP)

James Ostell-National Center for Bioinformatics, National Library of Medicine

Rochelle Small-National Institute of Deafness and Other Communication Disorders

SUMMARY REPORT
NEUROINFORMATICS
INTEROPERABILITY WORKSHOP
JANUARY 8, 1999
MARRIOTT HOTEL, POOKS HILL

Goal

To discuss and solicit opinions on the challenges and opportunities and next steps needed to advance the formation of a distributed set of interoperable databases in Neuroscience. The overarching thrust of this endeavor is to promote the creation within the Neuroscience community of systems for information management that communicate among databases and datasets across Neuroscience and other major relevant fields (e.g., genome, protein, etc.).

  1. Jim Ostell, drawing examples from sequence databases, reviewed some of the elements aiding interoperability, including:
    1. Shared Linkages
    2. Accession Numbers
    3. Standard Vocabulary
    4. Standard Nomenclature
    5. In data or real time
    6. Data Models
    7. formal versus informal
    8. inter-structural and -functional relationships at varying levels of analyses
    9. model interactions, and the importance of continued maintenance of relationships, computational components, and linkages for original observed data vs. the insufficiency for this requirement for data interpretation.

      The heuristic nature of scientific discovery employs both the initial descriptions of raw data and direct linkages, followed by hypothesis testing and theoretical development to further establish more complex parametric interrelationships. Scientific progress requires that the informational relationships be maintained accessible, at least locally, if not more broadly interoperable across all databases.

    10. Levels of interoperability
    11. Compatible schema (Linkages and Data Models)
    12. Compatible datasets (Defined Formats or interfaces such as CORBA)
    13. Portability among hardware and operating system platforms
  2. Concerns Discussed
    1. Maintenance of Databases
    2. Definition of a dataset/database
    3. Criteria for selection (inclusion vs. exclusion)
    4. Updating - How and when made?
    5. When to discard
    6. Provisional sets of domains (i.e., since data models impose boundaries)
    7. Compatibility and interfacing multi-platform databases and communities of federated databases
    8. Feasibility studies to establish reliability and cross-validate both datasets per se, and database interoperability
    9. How to remain contemporary with technological advancement
    10. Who does all of the above maintenance?
    11. How should it be supported?
    12. Scope of Support
    13. Investigator initiated
    14. Contract
    15. Cooperative Agreements
    16. Focused RFAs
    17. Neuroscience Data
    18. Select neuroscience data for inclusion
    19. Define data models
    20. Specify domains of investigation
    21. Prioritize goals (identification of critical national needs for specified areas of investigation and their broader objectives)
    22. Specify types of Neuroinformatics tools required
    23. Confirm data reliability and validity
    24. Establish equivalence canonical models
    25. Standards
    26. How to engage the Informatics Community
    27. What are the current career-tracks?
    28. Cross-training Issues (across the Informatics and Neuroscience fields)
    29. Differential salaries (in the private vs. the public academic sectors; resource developers vs. Neuroscience research investigators)
    30. Summer Workshops
    31. Keystone Meetings
    32. Gordon Conferences
    33. Annual Neuroscience meeting

Summary Issues

At the end of the day, the issues highlighted were:

  1. To identify and prioritize the database needs of the Neuroscience community and determine how they differ from those of other communities.
  2. To specify the informatics skills needed to aid Neuroscience and how to interest and recruit individuals with these skills.
  3. How to achieve interoperability?
Setting Priorities

Three major recommendations were made:

  1. Establish an Advisory Committee comprised of Informaticians and Neuroscientists.
  2. Establish mechanisms to develop common or compatible nomenclature, data models, and organizing themes, and encourage or mandate interoperability components to neuroinformatics projects.
  3. Focus initially upon Neuroanatomy and establish a nomenclature, data models, and an organizing theme.

Additional Post Meeting Comments:

Encourage or mandate interoperability components to neuroinformatics projects. The focus should not be on neuroanatomy only.

There is a need to be more explicit about defining, from a functional point of view, what neuroscientists need from databases and where they see databases making a significant contribution to the way they do science. In the end, it may turn our to be a fragmented answer, with different specialties having very different needs. But it may be worth laying out the general landscape of imaging, neural circuitry, receptor/gene expression databases, etc. and tying these to how they can be used in conducting research as opposed to being simply compendiums of facts. Once this is laid out, it may be easier to answer questions such as comparing the needs of neuroscience vs. other communities and issues of interoperability.

The critical next step involves the delineation of alternative Data Set strategies to begin encompassing the scope of the project problem and exemplars of solutions. For example, one potential approach might involve nigrostriatal gene expression-dopamine structure and function (etc.)-intracellular cascades-relevant motor programs-striatal memory-Parkinson's disease-L-DOPA-pallidotomy-deep brain stimulation....... etc. An alternative would be the use of a relatively well-defined system, such as the hippocampal system, with analysis from single channel to synaptic transmission to cell system and memory, entering data from LTP to chemical messengers to patient H.M.While we could formulate many different examples, the point, I think is made that different categories and strategies must be articulated for choices to be made for the direction of the overall project.

Undoubtedly an additional meeting would be required to consider the merits of hierarchical, structural, functional, systems, disease-oriented models as starting points.