/home/precedings/apps/precedings/shared/storage/document/2008/1983/npre20081983-1

[ ]

1st INCF Workshop on Sustainability of Neuroscience Databases

December 13-14, 2007
International Neuroinformatics Coordinating Facility Secretariat
Stockholm, Sweden

Authors

Jack Van Horn and Jaap van Pelt

Scientific Organizer

Jaap van Pelt, VU University Amsterdam, The Netherlands

Workshop Participants

Jan G. Bjaalie, INCF Secretariat, Stockholm, Sweden
Chris Emblow, Akvaplan-niva AS, Polar Environmental Centre, Tromsø, Norway
Sten Grillner, Karolinska Institutet, Stockholm, Sweden
Jack Van Horn, UCLA School of Medicine, Los Angeles, USA (Rapporteur)
Tadashi Isa, National Institute for Physiological Sciences, Okazaki, Japan
Martin Kersten, Center for Mathematics and Informatics, Amsterdam, The Netherlands
Wouter Los, University of Amsterdam, Amsterdam, The Netherlands
Alessandro Orro, CNR Institute of Biomedical Technologies, Segrate, Italy
Roman Mou

cek, University of West Bohemia in Pilsen, Plzen, Czech Republic

Martin Nawrot, Free University Berlin, Berlin, Germany
Matias Palva, University of Helsinki, Finland
Jaap van Pelt, VU University Amsterdam, The Netherlands
Fiona Reddington, NCRI Informatics Initiative, London, UK
Shankar Subramaniam, University of California at San Diego, La Jolla, USA
Jostein Kandal Sundet, Oslo Central University Computing Center, Oslo, Norway
Shiro Usui, RIKEN, BSI, Wako, Japan
Ilya Zaslavski, University of California, San Diego, USA

Supported by the EU Special Support Action INCF, the INCF Central Fund, and the Swedish
Foundation for Strategic Research

[ ]

1. Executive Summary

Neuroscience databases broaden and extend the scope of both published and non-published data by making them widely available

to focused or open communities. Data organized in these fashions are searchable, viewable, and suitable for secondary explora-

tion far beyond the purpose of their original collection. Governmental scientific agencies encourage the development and use of

these resources but are not interested in long term support, per se. Once a database has been developed, questions arise as to who

benefits from the resource, how it is being maintained, what data model underlies its organization, and whether it is interoperable

with other resources. Failure in any of these areas may mean that the database cannot be sufficiently sustained as a resource. The

sustainability of these endeavors is of paramount importance to the field of neuroscience and the INCF is uniquely positioned to

examine how databases can maximize their long term sustainability, attract users, and provide linkages to other data resources in

such a way as to make each one an indispensable component of a larger whole. This meeting sought to explore these issues and

make recommendations for the INCF to pursue that would involve rating, ranking, and supporting database sustainability.

2. Introduction

Neuroscience databases provide a diverse collection of com-

munities with access to raw and meta- data, analytical tools

and computational models for use in new research, methods

development, and science education. The development of

these valuable resources often requires several years of active

and focused effort in order to meet the needs of a particular

research collaborator or to be able to be used by the neurosci-

ence research community. Building the database architecture,

populating it with content, and linking that content to content

residing in other internet-enabled resources increases the over-

all richness of those resources. The mining, re-analyses, and

visualization of data from archives of data from previously

published research articles enables independent validation,

new methods development, use in education, and novel neu-

roscience outcomes.

Nevertheless, how these databases sustain their activities in

the long term remains a question without a clear or satisfac-

tory solution. Many variables come into play when consider-

ing database sustainability, not the least of which is ongoing

governmental support for database curation and tools develop-

ment. Others include an engaged process of curation, systems

support, and active scholarly activity that draws from the re-

source. Moreover, the fact that people actually use these re-

sources to conduct novel scientific discovery, and have come

to rely on them, means that should they fail to maintain their

sustainability, a certain segment of the neuroscience research

enterprise (not just the database in question) will be affected.

In addition, their usefulness in the training of the next gen-

eration of neuroscientists cannot be overstated. Government

funding agencies must examine carefully the impact of getting

such programs started, what it will take to continue their mo-

mentum following their initial construction, and who will be

inconvenienced should they falter. Organizations such as the

International Neuroinformatics Coordinating Facility (INCF)

are well positioned to monitor international database activities,

usage, and longevity.

In order to articulate its role in database sustainability, the

INCF organized the 1st INCF Workshop on Neuroscience Da-

tabase Sustainability at the INCF Secretariat, at the Karolinska

Institutet in Stockholm, Sweden from 13-14 December, 2007.

The goal of the workshop was to discuss issues related to the

sustainability of neuroscience databases, identify problems and

discuss solutions or approaches to these problems, and formu-

late recommendations to the INCF. Expert researchers were

invited to the workshop from the international neuroinformat-

ics community, as well as other disciplines within biology and

clinical medicine where sustainability issues have already been

approached and experience obtained.

[ ]

3. Sustainability Issues

 Summaries from Speaker

Presentations

Each of the invited participants framed the issues of sustain-

ability from his or her own unique perspective. Collectively,

several general themes emerged: the pervasive role that data-

bases are now taking in each domain of neuroscience; their

linkage with peer-reviewed publication; the role of govern-

mental funding agencies; the needs for minimal but useful and

interoperable standards for data and meta-data file exchange;

the needs for business models; the instruments for quality con-

trol, and the role of the community and other stakeholders in

the endeavor of databases and data sharing--those contribut-

ing to and those drawing from databases. In the brief sections

that follow, we outline the main points raised by each speaker.

I. Sten Grillner

Databases are important infrastructures, unique and competi-

tive. They deserve commitments by funding agencies for their

support, through regular reapplication for renewal and conti-

nuity. It is recommended that funding agencies allocate funds

for their long-term continuity as they are doing, for example,

for funding agency-associated research institutes, or interna-

tional programs such as particle accelerator facilities (CERN),

biodiversity programs (GBIF), or networks of astronomical

observatories. Promotion of standards for data sharing will

increase community involvement and use, resulting in finan-

cial benefits. INCF shall support and take action in web-based

community interaction of databases of data, tools and models.

II. Jaap van Pelt Sustainability Issues of Neuroscience

Databases

The long-term success of neuroscience databases depends on

a variety of factors:

· Sustainability of the database technical facility: The fa-

cility must implement interoperability standards (e.g.,

ontologies and benchmarks), follow technological de-

velopments through regular upgrades, and be designed

(scale, granularity, central/distributed structure) to ac-

commodate limited lifetime of platforms.

· Sustainability of database content: The scientific qual-

ity of database content must be maintained at the high-

est level, through instruments such as regular quality as-

sessments, carefully evaluating the lifetime of the data

and metadata, regular curation of the data, and protec-

tion against database degradation. The owner of the data

has particular responsibility in this ongoing process. The

quality issue also concerns tools for analysis, visualiza-

tion, and modeling.

· Sustainability of database operation: The facility requires

technical staff, maintenance, and user support. Data pro-

tection and security require appropriate measures for

access rights, copyrights, and detection of misuse and

unauthorized use. The facility should optimally meet the

needs and requirements of the users and be integrated in

scientific practice. Investments are needed in training,

promotion and visibility, user friendliness, and documen-

tation, ease of (meta-) data entry/retrieval. User groups

must be involved in evaluations through measures of use,

applicability, and needs. IPR and ethical issues need to be

implemented as well as journal policies.

· Sustainability of database enabling resources: Organi-

zations can have particular roles in providing enabling

resources for the data base facilities such as represent-

ing user groups (Societies, Journals), application areas

(Companies), support and implementation (Institutions,

Funding Agencies). Appropriate financial resources are

crucial to cover the costs of database accommodation,

equipment, management and operation.

III. Shankar Subramaniam Interoperability and Data

Integration in Neuroscience Databases

Interoperability of databases sets requirements on data orga-

nization and presentation (ontologies, databases, query tools,

interfaces), data interoperability (relational rules, data formats,

disambiguation) and data integration (heterogeneous data in-

tegration, legacy knowledge integration, weighing knowledge

association), data analysis (statistical tools, integration with bi-

ology, reducing complexity, models), and data curation. These

Neuroscience Infrastructures are even more difficult than those

[ 7 ]

in bioinformatics because of the curse of multiple dimensions

(anatomy and atlases, models), the modality of measurements,

the granularity of description, and shape and topology issues.

Available databases may be distinguished from those contain-

ing neuroscience-related experimental data that are publicly

accessible; those containing neuroscience knowledge, data-

bases of tools and tool registries, or links to neuroscience in-

formation portals (e.g., Human Brain Project, BIRN, BrainInfo

Sites).

Recommendations:

· Facilitate sharing of data and databases: INCF should en-

courage interoperable formats for data representation and

presentation. The use of standard ontologies is recom-

mended.

·Provide tools for interoperability between data and da-

tabases: INCF should encourage creation of data object

porting capabilities (using markup languages) and me-

diators for interoperable querying of data between data-

bases. INCF should encourage query infrastructures with

APIs that will facilitate modular usability across diverse

data.

· Make interfaces interoperable and uniform: INCF should

encourage development of portable visualization inter-

faces (java applications) that can be used in modular and

portable manner.

·Promote data integration in neuroscience: Given the im-

portance of anatomy, INCF should encourage the use of

standard brain atlases and mapping tools that will facili-

tate mapping of diverse data--molecular, cellular, tissue

images--onto anatomy. This will provide a three-dimen-

sional perspective of the data mapped to function.

·Emphasize the role of models: INCF should encourage

the presentation of models in neuroscience in portable

markup language formats that will make it easy to map

models to other data. Also, annotation of models needs to

be an integral part of the models.

·Encourage curation by community: INCF should create

a global curation community using the auspices of jour-

nals in neuroscience/informatics for validation of data,

models and anatomical correlates. INCF should encour-

age integration with the neuroscience clinical community

to deal with pathology associated with the brain.

IV. Shiro Usui J-node Sustainability Scheme Including

Government Support

Sustainability of the INCF Japan-Node is supported by a well-

organized scheme of coordination, management, promotion,

and user and researcher involvement. This high level of orga-

nization also applies to the series of neuroscience topical and

focused brain science platforms under governmental support.

The Xoops scalable content management system XooNIps pro-

vides the technical support with tree-like links (

http://xoonips.

sourceforge.jp/

) as an open source.

V. Tadashi Isa National Brain Research Project

"Integrating Brain Research" from the Database Committee

Point of View

The Integrative Brain Research (IBR) Project Database dis-

tinguishes a top-down type (with neuroscientists and research

outcomes), a bottom-up type (with a neuroscientists social net-

working service (IBR-SNS), and a mouse brain-behavior phe-

notype database. Sustainability issues concern the questions of

how to get more data and more detailed information, how to

encourage incentives for uploading the data from individual

scientists, how to collect raw data for open access, and how to

continue after the research term of the IBR project. Actions are

planned to transfer the neuroscientists database and research

outcome database to the XooNIPS @ NIJC in RIKEN BSI, and

the SNS neuroscientists to some succeeding project of IBR,

while collaboration should be maintained with the NIJC and

Japan Neuroscience Society. The mouse phenotype database

should be maintained in conjunction with the NIJC.

[ ]

VI. Jack Van Horn Business Models for Neuroscience

Database Sustainability

Business models for neuroscience database sustainability need

to specify their value elements. These include need and audi-

ence (see example of bioinformatics), models (is the database

model particularly novel, general, or useful?), usage (does the

data resource get used and in what way?), community (who

does the database serve and how broadly?), literature (is the

data supported by peer-reviewed publications and/or represen-

tative of the whole field?), sustainability (by what means are

its ongoing activities supported?). Crucial for sustainability are

the support models. Examples are databases driven by consortia

(such as BIRN requiring membership, ICBM and caBIGTM), or

databases supported by consortia such as LONI, providing da-

tabase and computational services to multiple consortia-driven

projects (e.g., ADNI and BIRN) and individual labs (requiring

collaborative agreements with preference for funded research).

Such databases serve the needs of the members of the consor-

tia, but may also serve larger communities with tuned access

modes. An issue is what happens to a database when its consor-

tium completes its mission.

Databases must be tied to scientific literature and based on

agreed upon "classes of domain-specific metadata". Published

articles appear to report methods with increasingly reduced de-

tail, often relegated to "supplementary documentation", result-

ing in uncertainty on what was done and few if any replications

(e.g., neuroimaging). No agreed upon guidelines, format, or

framework exist for what should be provided in supplemental

documentation. The core set of methods information needs to

be identified and it should be expected to be fully disclosed in

peer-reviewed articles. An example is the Minimum Informa-

tion for Neuroimaging Description and Specification (MINDS)

(Appendix A1).

Questions concerning sustainability of databases include their

linkage to peer-reviewed research, their usage, their impact on

generating new research (new papers from old data?), govern-

mental support which is needed in the case of required depo-

sition policies, society sanction (could SfN, INCF, or others

sanction databases?), and level of service (basic or enhanced--

dependent on free or paid access).

Recommendations:

·INCF can help databases clearly define the target au-

dience benefiting from the resource and why there is a

need.

·Encourage DB developers to be open, candid, and realis-

tic about the data models.

·Showcase where and how data from the resource have

been used and be prepared to back it up.

· Advocate professional society and organizational sanc-

tioning.

· Encourage that database content be tied to the peer-re-

viewed literature whenever possible.

·Be vocal about governmental agencies making possible

long-term options for funding coupled to level-of-scale

user fee structure.

· Help database developers envision a clear "event hori-

zon" for when it may be time to stop curating new data

or for letting the database fade away gracefully should it

be deemed necessary.

VII. Chris Emblow The Society for the Management of

European Biodiversity Data and its Role in the Sustainability

of Taxonomic Checklist Databases

The Society for the Management of European Biodiversity

Data plays an important role in the sustainability of taxonomic

checklist databases. Experiences were obtained with the Euro-

pean Register of Marine Species (ERMS), covering four data-

bases: (1) the register of species (over 30,000 marine taxa), (2)

the bibliography of identification guides (840 publications),

(3) the register of species identification/ taxonomic experts

(600 individuals from 37 countries), and (4) the register of

marine reference collections, which were all completed with

the unpaid help of many individual specialists of Marine Spe-

cies. This was made possible by specifying agreements from

the contributors (i) to voluntarily provide data, information,

opinion, or other expert assistance to the ERMS project, (ii)

to retain the right to use and publish any data and intellectual

property created by the contributor, (iii) to authorize the proj-

ect to store, compile, modify, and disseminate data provided

[ ]

and derived by any means (e.g., electronic, World Wide Web,

book), and (iv) to recognize that products of the ERMS are

the copyright of the project and will not disseminate further

ERMS publications or data without prior permission of the

project Steering Committee. In addition, the project agreed to

(i) acknowledge the contribution of the contributor in publica-

tions of the ERMS, (ii) provide the contributor with a copy of

the ERMS publications, (iii) establish a new organization to

manage the ERMS after the completion of the project on 31st

March 2000, (iv) transfer all ownership of data and intellectual

property collected as part of this project to the new organiza-

tion by 31st March 2000, and (v) to ensure that the contributor

has the right to elect individuals to the management committee

of this new organization.

The Society for the Management of European Biodiversity

Data (SMEBD) was established in 2000 as a "not-for-profit"

company limited by guarantee and not having a share capital

based in Ireland, and with contributors agreements transferred

to SMEBD at the end of the project. Its role in sustainability

was specified as (i) to act on behalf of its members to man-

age European Biodiversity data, including the European Reg-

ister of Marine Species, (ii) to provide a legal basis for the

protection of the members' contributed data, (iii) to facilitate

communication and interaction between persons interested in

biodiversity and its application to environmental management,

(iv) to promote the publication and dissemination of informa-

tion related to biodiversity, (v) to facilitate access to specialist

knowledge and scientific opinion on biodiversity, and (vi) to

raise funds to further the aims of the society.

In 2004, the Marine Biodiversity and Ecosystem Functioning

was initiated as the first EU FP6 Network of Excellence with

over 600 scientists, 56 member institutes, 36 associate mem-

ber institutes, and countries. As a new instrument of FP it

provided appropriate funding to maintain and update ERMS.

Further long-term commitment to host data was obtained from

Flanders Marine Institute (VLIZ) in ERMS 2.0.

VIII. Wouter Los Maintenance, Sustainability and

Management of Databases in the Environmental Sciences

For data maintenance, management, and sustainability,

ownership of distributed data is essential, as is the need to

organize different expert communities (at the disciplinary,

institutional and regional level). Data integration leads to

peer-reviewed authority files.

Data e-infrastructure, including interoperability with other da-

tabases and applications, is important to support the data flows

according to agreed standards. Biodiversity databases typically

contain complex data records of diverse and heterogeneous

data. Data are very often generated at distributed (monitoring

and collection) sites and may depend on human interpretation.

With parallel data taxonomies, data integration and data access

require regional and global cooperation.

Recommendations:
Business plans for databases and services:
Increase and organize the ownership of databases

· Consider experts contributing as (co-)authors

· Organize peer reviews

· Track and publish ownership of data to provide proper

credits

· Adopt Global Unique Identifiers (GUIDs) to trace uses

· Formalize institutional commitment

Databases are an asset: exploit these accordingly

· Provide mission statements that express such awareness

· Implement common standards and data management

practices

· Promote actively international data sharing

· Develop/contribute to new services and products

[ 10 ]

Organize the components of the database
development, maintenance, etc.
Infrastructure development

· Benefit from the potential of INCF

· Design the stratification and allocation of responsibilities

in the chain of data flows

· Promote new infrastructure capabilities

IX. Fiona Reddington Multi-Disciplinary Data Sharing: A

UK Perspective

What is seen as the informatics problem can also be consid-

ered a consequence of success. In databasing and data sharing,

one is faced with different types of data, large data volumes,

different data standards, inaccessibility of much research data,

multiple stakeholders, varying (dis-)incentives and willingness

of researchers to share their data, lots of activities while little

coordination, and little funding for standards, education or in-

frastructure projects.

In order to address these issues the UK National Cancer Re-

search Institute (NCRI) initiated a partnership and developed

a data sharing policy that could be a model for other disease-

related data. This policy is shared among the partners of NCRI

and builds on existing resources and standards, recognizing

that this will be easier in some areas than others, leading to

phased implementation. It empowers the research community

and recognizes resource requirements.

NCRI can help identify and promote standards (maintaining

and promoting a "planning matrix"), deliver infrastructure to

support data sharing, and support demonstrator projects and

engage in/foster debate on key topics (such as hosting work-

shops, publications and conferences, presentations at external

events).

Future actions focus on (i) the promotion of national and in-

ternational collaboration (by linking the UK cancer research

community, unifying cancer and non-cancer research teams,

irrespective of size, frees up researchers to focus more on the

science, integration of connected resources globally, and build-

ing a flexible but consistent infrastructure for future sources/

services ), and (ii) fostering a data-sharing culture (by making

resources more accessible, reusing resources (data and servic-

es) across multiple domains, reducing duplication thus saving

time and effort, and adopting data standards and data sharing.

Summarizing, the NCRI Informatics Initiative is helping to

work with the UK and international community to build reusa-

ble infrastructure, and to change culture to make the infrastruc-

ture usable. NCRI anticipates that its approach, being open and

voluntary, is in everyone's best interest.

Recommendations:

·Focus on communication

·Share responsibilities with stakeholders

·Involve end users

·link NCRI with INCF and other international collabora-

tors

X. Jostein Sundet Data Exchange in International

Collaborations

Success in data exchange in international collaborations

depends crucially on the services available for the

users. The research computing services offered at the

University of Oslo facilitate researchers in providing

access to (large) storage pools, in computation and

visualization. It is important that services are offered at

the IT-training and knowledge level of the researchers.

Challenges in data exchange include redundancy of data,

authentication, locality of data versus computational

resources, bandwidth dependencies, common ontologies,

useful toolboxes, publicly available data and results

(journals), and apparent community agreement. Support

is given to open-source policy and the philosophy that

"the more databases the better (particularly for rare

diseases)". University of Oslo offers storage services to

the CERN grid, to the Accent EU NoE, and to NorStore

as a collaboration between Universities. It also offers

a virtual research environment, mostly for document

exchange.

XI. Ilya Zaslavsky Long-Term Preservation of Spatial Infor-

mation: Research Issues and Supporting Infrastructure

According to Reagan Moore, preservation is an archival pro-

cess through which a digital entity is extracted from its cre-

ation environment and migrated to a preservation environment,

while maintaining authenticity and integrity information. The

extraction process requires insertion of support infrastructure

[ 11 ]

underneath the digital material and characterization of parame-

ters, such as the authenticity and integrity, the digital encoding

format, and the display operations. The goal is infrastructure

independence, i.e., the ability to use any storage system, data-

base, or access mechanism. In this sense, preservation is simi-

lar to information integration--but in the temporal domain.

A preservation strategy must provide emulation (to migrate the

display application onto new operating systems, equivalent to

forcing use of candlelight to look at 16th century documents),

transformative migration (to migrate the encoding format to

the new standard, migration period expected to be 5-10 years),

and a persistent object for the characterization of the encod-

ing format and the migration of the characterization forward

in time.

A preservation environment must include a digital library in-

frastructure that supports the preservation metadata, the ar-

rangement and description of items, and access mechanisms;

and a data grid infrastructure that supports shared collections

that are migrated forward in time, the management of technol-

ogy evolution, and administrative metadata providing status of

records.

In archiving and accessing spatial data, one is faced with:

· a variety of data types and models (open as well as pro-

prietary data formats and management systems, and sev-

eral emerging XML encoding standards), creating a need

for some level of unification on a non-proprietary basis,

· different spatial registration mechanisms (e.g., spatial co-

ordinates in a common (or well-described) system, labels

in an agreed upon (or well specified) ontology, location

relative to well-defined features, representative loca-

tion--especially when navigating across scales),

· data structures with different amounts of "intelligence",

· the need of encoding data quality (spatial, semantic) de-

pendent on the instruments, feature types, and transfor-

mations used,

· the need to specify metadata (the context, hyper-linked

content, time stamps, etc.),

· the preservation of "look and feel" and

· graphical user interfaces for data integration, archive cu-

ration, and the retrieval and transfer of archived data into

an operational environment.

As an example, the Smart Atlas is a grid-based Geographic

Information System tool for spatial integration of multi-scale

distributed brain data. It uses ontologies to delineate anatomic

features, disambiguate label assignments, link features, and

synchronize positions. The Smart Atlas tool is currently be-

ing developed by the BIRN-CC Data Mediation Team (

www.

nbirn.net/

The following checklist of issues is recommended for archiving

and preservation in the neurosciences:

·What are observation data preservation needs and chal-

lenges faced by neuroscientists?

·What are the appropriate archival forms for typical da-

tasets in neuroscience? To what extent can available re-

lational schemata, etc., be used as preservation formats,

how can these approaches be integrated with existing

standards such as the OAI (Open Archives Initiative)

standards and protocols.

·How can archival and observations metadata be harmo-

nized? The sharing and use of digital information in a

community is different from the information and pro-

cessing required for archiving in a long-term preserva-

tion environment. We need to incorporate the metadata

standards adopted in the archival community without

causing an undue burden on functioning systems.

· What are potential types of semantic mismatches, and

controlled vocabularies and semantic reconciliation/me-

diation tools needed to support long-term archiving of

neuroscience data? The neuroscience and archiving com-

munities have different notions for some of the basic

terms (e.g., a preservation record as understood by archi-

vists may not match the notion of record as understood

by domain scientists).

· What are the appraisal, accession, arrangement, descrip-

tion, preservation and access procedures to be imple-

mented for neuroscience data? What is appraisal value of

neuroscience records (and hence retention policies)?

·Once an archive is established, what is the minimal set of

queries and instantiation requests it shall support? (and

access restrictions)

[ 12 ]

·What are user interfaces for curating, updating, access-

ing, annotating and analyzing neuroscience data archives,

appropriate for both archivists and domain scientists.

· What additional analytical services and navigation tools

shall be implemented over large data archives?

· What is the software architecture of a neuroscience data

preservation testbed, and infrastructure for sustainable

distributed archiving of neuroscience data? Infrastruc-

ture-independence (to ensure that the archives can be mi-

grated to new media, platforms and data formats as they

become common) and replication of archives at different

locations must be considered.

· What are institutional arrangements and "governance

patterns" needed for a successful long-term archiving

testbed implementation? Specifically, what steps are

needed for integrating a preservation environment in an

operational neuroscience data collection/dissemination

system in a manner that establishes the archive as a trust-

worthy resource?

·What are the possible avenues for sustainability of a long

term repository? (What is the business model?)

Recommendations:

·Engage with communities that deal with long-term pres-

ervation professionally.

·Have a "preservation testbed" to develop answers to the

initial set of issues.

·Establish use cases for preservation.

XII. Matias Palva Data Format and Interface Issues for

Sustaining Multi-Scale Databases

There is a need for process and data management tools to

solve problems encountered in current data analysis workflow.

These problems arise from a combination of factors, such as

the many sources for neuroscience data (such as MEG, EEG,

EOG, EMG; eye gaze tracking; MRI, fMRI; psychophysics;

quantitative trait loci (QTLs), genotype data, twins; neuro-

physiology; all resulting in a high dimensionality of the data);

the manual work; the many operating systems, programs, and

formats; the technical and programming skills required for the

analysis process, and the incomplete documentation of process

stages and details. As a consequence, processes are slow, cum-

bersome and irreversible, with irreproducible outcomes, waste

of human resources, data integration challenges, and low re-

turn-of-payment from hardware investments. Data analysis

becomes a bottleneck for productivity and quality within a hy-

pothesis-driven science.

The platform developed by the OmniLyze Project aimed at

providing generic solutions for data and process management,

by developing a unified software system with graphical UI,

centralized storage, services for computing, data management,

process management, and entry through remote desktops or

the web. This approach improves quality and productivity,

decreases risks, facilitates data integration, makes the process

reproducible and publishable, and improves human resource

yield and hardware yield. In the view of the OmniLyze Project,

process and data management should become a commonplace,

everyday practise integrated into every step of the scientific

workflow. Every form of data storage is considered to be a

"database", and every database should offer data and process

management. Databases exist on many scales such as personal,

research groups, research centers, national, and global.

Sustainability of databases demands resources and continuity

in both usage and development. Generic database solutions

may be advantageous when resource income is proportional

to the number of users, but require data to be transportable be-

tween databases across scales and functions, with standardized

data formats and interfaces.

A standard data format (SDF) includes unambiguous defini-

tions for legacy and metadata. It would promote database sus-

tainability by user attraction, common UIs and usage, data and

code reuse, sharing, modularization, co-development, avoid

the need for specific database-database interfaces, support for

database continuity, and decreased development and mainte-

nance costs.

Recommendations:

A roadmap for INCF for data format and interface issues was

suggested to include the following actions: (i) INCF arranges

groups of experts to compile existing data and metadata defini-

tions and format conventions, (ii) workgroups propose novel

or amalgam definitions and formats, (iii) INCF compiles these

and declares an SDF, (iv) database and equipment developers

can produce the specific interfaces, and (v) INCF provides a

portal to SDF resources and tools. SDF would be a basis for

standardized process definitions, APIs, interfaces, and meta-

interfaces.

[ 13 ]

At a technical level it was proposed to consider the HDF (Hi-

erarchical Data Format; http://hdf.ncsa.uiuc.edu) as an attrac-

tive candidate to carry the role of SDF in neuroinformatics,

as it handles data of any type, complexity, or size, it supports

hierarchical grouping of high-dimensional datasets, and it has

an efficient API providing fast data access including parallel

I/O. It is usable in many computational platforms (AIX, HP-

UX, IRIX, Linux, SunOS, Windows, Mac OS), is stable and

sustained, widely used (EOS, 3 Pb, 1.6 M users) and is open

source, costing nothing.

XIII. Martin Kersten Scientific Databases: The Story

Behind the Scenes

Key issues in the research of efficient and effective database

technologies are the ultimate (virtual) machine architecture for

database processing and the challenge of blending information

retrieval with large-scale database processing. Thinking about

neuroscience databases starts with an initial (data) exploration

of the field followed by a series of stepping stones marking (1)

the multimedia dimension, () the geometric dimension, ()

the lineage dimension, (4) heterogeneous databases, (5) GRID

databases, and () the semantic search.

Stepping stone 1: Multimedia databases typically contain

large volumes (>Tbyte, >Pbyte) of raw data with partition-

ing based on image, video segmentation, and indexing based

on feature vectors. Query challenges concern proximity and

probability based search, are CPU intensive with user-defined

predicates and content-based information retrieval. For exam-

ple, the database

http://monetdb.cwi.nl/demo/bond

consists of

100,000 images, with 25 patches extracted from each image,

and a 14-dimensional feature vector derived for each patch,

resulting in 2,500,000 images. The challenge is to find similar

images based on Euclidian distance with sub-second response

time. A novel database algorithm was developed to solve K-

nearest neighbours (k-NN) search. An alternative scheme is

to determine the probability that an image can be generated

with a limited number of Gaussian mixtures. To this end, fix a

limited number of Gaussian Mixture Models (GMMs), use an

Expectation Maximization algorithm to fit the model over the

image, and search similar images by comparison of the GMM

model parameters.

Stepping stone 2: Any geometric abstraction of reality pro-

vides a good navigational map. Database storage and indexing

support for 2D is mature with R-trees and Quad-trees, but com-

mercial database vendors "do not like them". Open research is-

sue is to support D query embedding and scaling out towards

3 and 4 dimensions and temporal support. This approach is

researched extensively in Geographical Information Systems.

A lesson drawn from it is to avoid abundance of reference

models.

Stepping stone 3: To ensure data lineage is a problem encoun-

tered in many scientific databases, namely the ability to travel

back in time to understand, redo and judge the derivations.

Challenges here are: (i) how to keep track of the complete con-

text with data, software, parameter settings, etc., (ii) how to

redo part of the analysis, and (iii) how to store and remember

the lineage trails. An example is the AstroWise project in Gro-

ningen that keeps track of a complete workflow for telescope

data analysis in a large Oracle database. All derivations are -

line python programs.

Stepping stone 4: Sharing of heterogeneous information is the

key problem in many databases. The standard approach here

is to use commonly approved vocabulary and standard syn-

tax, with XML being the standard language for self-descriptive

data and its exchange between software systems. The data-

base community is actively working on improving XML itself

as well as XML-based tools and applications such as XQuery

and Xupdate database engines, but it is not easy! The chal-

lenge is how to scale to large XML stores, how to efficiently

search components, and how to realize structural information

retrieval.

Stepping stone 5: GRID technology is partly a re-invention

of distributed database technology, distributed programming,

and high-performance computing with a focus on Authentica-

tion-Authorization-Access and data shipping over wide-area

networks. Data distribution, replication, and parallel query

processing has been well studied over the last three decades.

World-spanning application deployment is centered around

web-services, the middleware layers to publish and integrate

different sources.

Stepping stone 6: Ontology integration is one of the most

pressing challenges for the semantic web to take off, but in-

tegration of technology with databases is still immature. RDF

and OWL are the leading paradigms; SPARQL is the first at-

tempt to bridge the gap between traditional database manage-

ment and semantic web technology. The integration of the

cultural heritage field at

http://e-culture.multimedian.nl/demo/

may serve as an example.

[ 14 ]

Recommendations:

· Manage sizeable scientific databases using open-source

database technology in order to capitalize and steer ex-

pertise development

· Articulate the technical challenges in concrete require-

ments in order to know what you have to provide

· Benefit from progress in semantic web technology that

might provide a good stepping stone requiring the devel-

opment of ontologies and semantic bridges

· Make a dual community portal to mobilize the world,

e.g.,

http://cas.sdss.org/dr/en/

· Team up and make it a success--crossing the great di-

vide between neuroscience and computer science; this

is challenging and rewarding if and only if the building

of the bridge starts from both ends, and parties recog-

nize and respect each other's core business. The database

community can provide knowledge on modeling, query

processing, algorithms, data structures, scalability, per-

sistency and flexible database systems.

Refs:

http://monetdb.cwi.nl/

http://monetdb.sf.net/

http://

www.monetdb.com

XIV. Roman Moucek EEG and ERP Records and

Processing

Data in EEG and ERP research originates from multiple elec-

trodes and carries contextual information. A simple and flexible

format for the exchange and storage of multichannel biologi-

cal and physical continuous signals is provided by the Euro-

pean Data Format (EDF). Published in 1992 (Electroenceph.

Clin Neurophysiol. 82: 391-393), it has become a standard for

EEG recordings in commercial equipment and multicenter re-

search projects. An extended version European Data Format

(EDF+) was published in 2003 in Clin. Neurophysiol. 114:

1755-1761.

Recommendations:

· Work on data sharing in specific domains.

· Find the authorities, domain experts and technical ex-

perts.

· evelop domain standards (ontology, metadata), technical

standards (file formats, database structures), technolo-

gies (programming languages, distribution, replication,

encryption), and interfaces. A question is how to find the

right level of domain differentiation.

XV. Martin Nawrot Three Challenges: Data Complexity,

Data Format Confusion, and Non-transparency of Data Anal-

yses

Exchanging and sharing data is important for scientific coop-

eration. Problems arise from the complexity of the data and

the confusion from data formats. Commercial electrophysi-

ological equipment often provides proprietary (closed source)

file formats and analysis software with very limited "built-in"

data analysis and poor script languages. As a consequence the

experimenter is often "stuck" with his company/format, and

faces very limited data exchange across formats (format com-

munities). It leads to (expensive) in-house hacks for data ex-

port/import, making the process fragile, error prone and vul-

nerable (no docs, no compatibility).

Suggested measures are to define open-source "unified" file

formats, provide free services and expertise (for import / ex-

port functions, professional documentation, data base for data

storage/exchange, teaching and training), enforce open-source

standards, improve quality and flexibility (development com-

munity), and provide single interfaces to programming envi-

ronments (Matlab, Python,...). Questions arise as to whether

such a "unified data format" is realistic, as it supposes no com-

petition on the market. An intermediate step in electrophysiol-

ogy, adopted by a variety of companies, may be the application

programming interface (API) Neuroshare

http://neuroshare.

sourceforge.net

), with access to a variety of proprietary for-

mats. It is, however, platform specific (Windows only), de-

pendent on closed-source DLLs by industry, and no common

standard is proposed.

[ 15 ]

A third problem arises from the non-transparency of data anal-

yses. Data analysis links experiment and theory and high qual-

ity analyses and rigorous testing of theoretic models are crucial

for the scientific advancement. Published analyses are gener-

ally in-transparent, while sophisticated tools are not easily ac-

cessible. Commercial software packages and poor data access

are limiting factors. Suggested measures are to provide a code

repository for data analysis (open source, multi-lingual), a test-

ed toolbox for neural data analysis (open source), professional

support and documentation, and teaching and training.

XVI. Alessandro Orro Data Management in the EGEE Grid

Infrastructure: The BioinfoGRID Experience

Analysis of biological data often goes with computational in-

tensive tasks on huge amount of data obtained from custom

databases with indexed flat files or local data from local file

systems. A way to handle this is by using biological databases

organized in a Grid, a network of clusters. Main issues are the

installation and maintenance of the databases in the Grid En-

vironment, the optimization and scalability (i.e., minimizing

waiting time, storage and transfer costs), and the handling of

versions.

Recommendations:

· Make database for analysis mainly read-only

· Relational/Object (for LIMS, for interoperability) and

custom database (for analysis)

4. Workshop Discussions

Following the individual presentations of the expert partici-

pants, three discussion groups were formed to discuss sus-

tainability issues from the community, the technical and the

standards points of view, respectively. On the

day of the

workshop, a discussion group was formed on "additional is-

sues of sustainability". The next paragraphs summarize the

main outcomes of these group discussions.

4.1 Community Issues of Sustainability

Clearly define the community (e.g., who is the audience for

the resource?)

· Generators, Organizers, End Users (including clinicians),

Funders, Journals, Libraries, Others

· Identify roles and needs of each

Communicate

· Engage all stakeholders and articulate simple but long-

ranging goals

· Re-assure users outside of consortia that their needs are

also being considered

· Involve journals and editors in the process

· Raise the awareness that the enterprise of neuroscience is

bigger than any single sub-community

· Provide mechanisms for incorporating feedback on roles/

needs/wants from users (wiki pages, bulletin boards,

etc.)

Formulate (Standards)

· Develop focused but flexible standards

· Follow best practices where available (guideline docu-

ment)

· Make standards decisions open to stakeholders and com-

munity input

· Review regularly and when needed

[ 16 ]

Educate

· Lower barriers to community availability/access

· Operational transparency should be high and it should be

easy to get started with data sharing

· Empower next generation of neuroscientists by encour-

aging use of databases in dissertation meta-analyses and

quantitative literature reviews

Evaluate

· Develop and utilize unambiguous/unbiased database us-

age statistics

· Give credit to database developers

· Give credit to those who have shared their data (for pro-

fessional advancement)

4.2 Technical Issues Involved in

Sustainability

Limitations of the technical domain

· Technical issues are secondary to community and stan-

dards development

· Understanding how neuroscience community is orga-

nized and works with data is critical in developing infra-

structure for data sharing and sustainability

· To what extent shall INCF focus on technical solutions,

beyond cross-walks and bridges across frameworks

(metadata, ontological, spatial)? What would be an

"INCF stamp of technical approval" for data: Standards-

compliant (which standards?) with complete metadata,

Cross-walk-ready?

Query languages

· Still early to embrace newest query languages--typing

scheme of XQUERY is too complex and incompatible

with SQL. Data-centric is still the focal point (not docu-

ment-centric)--most of the data can be safely expressed

in relational schema.

· Need a comprehensive data model, integrating datasets,

documents and annotations.

· Isolate large neuroscience datasets: strategies for serving

and querying them are different (perhaps OLAP, image

services, etc.).

Analysis and benchmarking

· Target analysis environments: Matlab / Octave / R /

Python. Open source is preferable but Matlab is over-

whelming. There is a need to figure out use cases and

scenarios, as well as recognized benchmarking.

· INCF is recommended to actively promote/solicit open-

source solutions, develop case studies supporting open

source/comparisons with proprietary systems, develop

understandable benchmarks (and perhaps awards for

reaching such benchmarks).

Towards a complete solution

· A vision of a complete solution/scenario sees the task

broken down into manageable pieces, with well-defined

metrics and a mapping to recognized user/developer

roles, as well as to an infrastructure/service-oriented ar-

chitecture (SOA).

· Look at experience of NIST organizing advancement

of technology for informational retrieval. Define target

problems, explicate queries, and come up with indepen-

dent solutions to be compared at a workshop: This is a

potential model for INCF.

Grid and web services

· The task here is to survey the scope and opportunities

of neuroinformatics integration: find who or what tasks/

scenarios need integration, determine what sources are

likely to be queried within most scenarios, and support

interesting integration testbeds scenarios (testbedding

experience of OpenGIS may be relevant).

· Published service signatures are needed for web services:

some XML standard for output (BrainML or anything),

and web service wrapping tools for relational and O-O

sources.

INCF Data Portal and Annotation Environment

· INCF may work on a NeuroWiki of BrainWiki (off INCF

domain?), integrated with neuroscience publications/

journals. But who will be the editors and which social

networks will be addressed (off INCF)? Easy annotation

and voting/ranking tools are needed (But what is the an-

notation model?), as well as a centralized gateway to all

sorts of organized resources, neuroscience clearinghouse

(à la kdnuggets.com).

[ 17 ]

Long-term preservation

· Often used datasets could be replicated at the central site,

with centralized archival management/curation, and tools

for format conversion/migration/annotation. Neurosci-

ence data may have cultural heritage value, for instance,

future generations of researchers may be interested in

the first digital atlases of the brain, requiring long-term

preservation strategies for selected atlases, but how do

we select atlases and other neuroscience data to archive,

what is the archiving method, and where would future

researchers find such atlases? UNESCO may provide

resources for long-term preservation. This might be a

unique long-term role for INCF

Privacy/de-identification

· Recommendations need to be formulated on ethical and

patent/copyright issues, and de-identification require-

ments for integrated datasets.

· Technical issues include grid and web service security,

access control, single sign on, etc.

What can we learn from related communities?

· How are similar complexities managed in neighbor com-

munities?

· What are the limits of applicability of standard Service-

Oriented Architecture (SOA) solutions, and how do stan-

dards organizations function in other communities (e.g.,

OpenGIS in the geospatial field, emerging standards in

astrophysics, biodiversity)? And to what extent are these

experiences relevant? For instance, in GIS there were

several known vendors, and it was a matter of having

them support a few common exchange mechanisms--but

in neuroscience?

Policy to support sustainability

· INCF could identify the data resources with highest in-

formation value, and the interconnections between these

resources. Then, INCF can specify which resources shall

be preserved and at which schedule, which resources are

not sustained, and which resources have low information

value and do not need to be sustained.

4.3 Standards Issues of Sustainability

The motivation of the working group was to identify issues for

standardization, to provide a framework and rules for adopting

standards, to discuss the role of INCF vis-à-vis other Standard

Bodies and the question whether INCF should be a certifying

body for standards.

Standardization issues were identified as data, databases,

ontologies, tools, models, and interfaces.

For standards to be established they need to be adopted

by several groups/organizations/agencies, be portable across

systems, and be translatable into exchange formats. Standards

should have a minimal core but be extensible, accommodate

as many types of experimental hardware as feasible, and be

compatible with existing standards.

Standardization topics should cover a range of granulari-

ties: genomes and gene products (no need for developing new

standards), cells and tissues, networks and systems, anatomy

and images, brain regions and atlases, functional mapping, dis-

eases, dynamical data including development, and models.

Data  should have a markup language with metadata info for

formats, experimental information, granularity, description

of terminology, and minimal standards. It should be portable,

scalable and extensible, and needs an ontological framework

on which the data is based. Examples for such standards are:

· for genomes and genes: NCBI/EBI,

· for gene expression and proteomics: MIAME and HUPO

standards,

· for network data: NeuroML and extensions,

· for image data: DICOM,

· for anatomy: Standard atlases, Neural names.

A major issue is also to find common representations or cross-

mappings of representations for the different fields and conver-

gence of languages.

[ 18 ]

Databases  should be based on defined ontologies and sche-

mata that are portable (in visible formats). They should allow

for import/export of database data in exchange formats. Query

engines must be integral to databases and be defined explicitly.

Languages and source code specifications must be provided

for database applications.

Models and Tools  Models should be in exchange formats

with full specification, be portable, with defined granular-

ity, and input/output specifications. Conceptual and analyti-

cal models must be defined explicitly, minimally, and where

possible a computational model must be provided. Graphical

representations are recommended. Languages, compilers and

interfaces need to be specified for tools. Tools are needed for

format conversion that allow for interchanging between differ-

ent data formats.

Interfaces and APIs (standards) The Web is taken as a stan-

dard for interfaces (user interface). Each interface must have a

defined API, with specifications for graphical interfaces, porta-

bility, query, and use cases.

4.4 Additional Issues of Sustainability

Teaching and training applies to writing papers (all

components must be of high quality), to providing tools for

good laboratory practice (tools and standards), use of well-

designed databases and metadata, to process management,

collecting and sharing data, E-learning courses, practical

courses, exchange of students and lab visits, teaching resources

(lecture notes, course scripts, citable lecture notes with review

editorial boards), and taped lectures.
Software development Open-source code policy was seen as

an important issue. "Standardized" open source code / toolbox-

es should be developed by open-source toolbox communities,

and open-source web dissemination (e.g., via SourceForge).

INCF can be helpful in initiating these communities based on

already existing initiatives in member countries, (e.g., Xoo-

NIps under the Japan node, pilot project with Bernstein center

projects and CARMEN for electrophysiology, other communi-

ties for image analysis toolboxes, for data mining and machine

learning (Weka open source in Java; http://www.cs.waikato.

ac.nz/ml/weka/), and for signal analysis tools). Examples of

open-source tool boxes are Numerical Recipes, OpenG (the

LabVIEW open-source community), Python/SciPy and R.

Toolbox communities can also support convergence of meta-

data descriptions (at all their levels). Adherence to standards

(ISO) is important. A tracking system was suggested for au-

thors/papers having used particular data.

5. Recommendations

Following the plenary presentations of the discussion group

reports at the second day of the workshop, the participants dis-

cussed and outlined a series of recommendations to the INCF

of actions to support sustainability of neuroscience databases.

i. INCF establishes a moderated web-based
infrastructure with specific issues for discussion by
the community

This infrastructure will enable community discussions and

documentation on choices of data, databases, etc. for minimal

information standards, discussion of models for broad utiliza-

tion, and access to reference data, ontologies, atlases, markup

dictionaries, and other relevant objects.

ii.INCF engages peer-reviewed journals in the
process of identifying domain-specific minimal
information recommendations for the sharing and
sustainability of neuroscience data.

·Journals remain the principle means of scientific com-

munication and provide motivation for focused groups

of neuroscience researchers to collaboratively engage in

distilling minimal information required for data exchange

and utilization in defined domains of neuroscience. In

addition, journals have the ability to reach the largest

segment of the neuroscience research community. INCF

can engage the journals to seed the process of providing

domain expertise for minimal information standard defi-

nition to support experimental methods reporting which

is also in the interest of the journals themselves.

·INCF encourages journals to publish special issues/sec-

tion with articles discussing minimal standards for dis-

semination of neuroscience data, methods, tools, and

models.

[ 19 ]

iii. INCF identifies specific types of data/databases
and a set of researchers who are generating and
disseminating these data to form a special interest
group that will develop the minimal information
standards for that data/database.

· Neuroimages, microscopic images, electrophysiological

recordings, EEG/MEG, histology, and optical recordings

are examples of mature types of data that INCF can begin

to explore.

· INCF identifies and engages experts who disseminate

such data and who are well motivated towards commu-

nity oriented approaches.

·INCF develops a mechanism for accrediting and ac-

knowledging experts who contribute to the above objec-

tives of INCF.

·Process of development is transparent, public, and feed-

back is solicited and welcomed.

·Minimal information standards should have broadest ap-

plicability in order to avoid unnecessary granulation.

iv. INCF identifies specific types of models/tools
and a set of researchers who are generating and
disseminating these theoretical/computational
models to form a special interest group that will
develop the minimal information standards (in
appropriate exchange formats, I/O, GUIs, etc.) for
those models/tools.

·Network, anatomical, and disease models are examples

for INCF to begin to explore.

·INCF identifies and engages experts who disseminate

such models and who are well motivated toward com-

munity-oriented approaches.

·INCF develops a mechanism for accrediting and ac-

knowledging experts who contribute to the above objec-

tives of INCF.

·Process of development is transparent, public, and feed-

back is solicited and welcomed.

·Minimal information standards should have broadest ap-

plicability in order to avoid unnecessary granulation.

v. INCF investigates existing neuroscience data/
tools/models clearinghouses and examines how
they can engage in coordinating dissemination
activities

·NITRC, NIF, SfN, and various research consortia are

examples of data/tool/database clearinghouses for the

INCF to investigate.

·Examine how this is done in other science disciplines.

vi. INCF examines how to serve as an accreditation
body

·Develops the criteria and process for evaluating data/da-

tabases/models/tools.

·Possible mechanism is via application for accreditation

to the national nodes.

·Recommendations then passed to INCF Secretariat.

·Posted on web, certification.

·Informs journals and societies of DB accreditation so as

to help them know which resources meet/beat expecta-

tions for best practices.

vii. INCF can facilitate grass-roots recognition of
need for data/database sustainability

·Satellite meetings/workshops at international meetings

on database sustainability.

· Engage member-state funding agencies to extent possible

involvement.

[ 20 ]

Appendix A

A1 Databases/Portals/Other References Used in the Report

Neuroscience

NIF

http://neurogateway.org/

Neuroscience Information Framework

NDG

http://ndg.sfn.org/

Neuroscience database gateway

CARMEN

http://www.carmen.org.uk/

Code Analysis, Repository and Modelling for e- Neuroscience

NITRC

http://www.nitrc.org/

The Neuroimaging Informatics Tools and Resources Clearinghouse

ICBM

http://www.loni.ucla.edu/ICBM/

International Consortium for Brain Mapping

Neuroshare

http://neuroshare.sourceforge.net/index.shtml

Open data specifications and software for neurophysiology

Other areas

NCRI

http://www.cancerinformatics.org.uk/

National Cancer Research Institute Informatics Initiative

caBIGTM

https://cabig.nci.nih.gov/

 Cancer Biomedical Informatics GridTM

Norstore http://www.norstore.no/ Norwegian Storage Infrastructure

Biology Workbench

http://workbench.sdsc.edu

web-based tool for biologists with Sequence/structure Databanks

Microarray Resource

http://genome.ucsd.edu

Proteomics Resource

http://www.cellularsignaling.org

http://www.mitoproteome.org

Metabolomics Resource

http://www.lipidmaps.org

Cellular Pathways

http://www.cytoscape.org

http://www.biopathwaysworkbench.org

Modeling Tools Statistical Tools

http://www.modelingworkbench.org

MonetDB

http://monetdb.cwi.nl

/ - open-source database system for high-performance applications in data mining, OLAP,

GIS, XML Query, text and multimedia retrieval

SkyServer

http://cas.sdss.org/dr/en/

- Sloan Digital Sky Survey

EGEE

http://www.eu-egee.org/

Enabling Grids for E-sciencE

BioinfoGRID

http://www.bioinfogrid.eu/

Bioinformatics Grid Application for life science

SourceForge

http://web.sourceforge.com/

global technology community's hub for information exchange, open-source soft-

ware distribution and services

Weka

http://www.cs.waikato.ac.nz/ml/weka/

collection of machine learning algorithms for data mining tasks (open source

written in Java, issued under the GNU General Public License)

Biodiversity Websites and Links

SMEBD

http://www.smebd.eu

Society for the Management of European Biodiversity Data

MarBEF

http://www.marbef.org

network of Excellence & ERMS 2.0

Fauna Europaea

http://www.faunaeur.org

ERMS

http://www.marbef.org/data/erms

The European Register of Marine Species (ERMS)

EurOBIS

http://www.marbef.org/data/eurobis

European Node of OBIS

MedOBIS

http://www.medobis.org/

Regional Repository of Marine Biodiversity Data

OBIS

http://www.iobis.org/

Ocean Bibliographic Information System

COML

http://www.coml.org/

Census of Marine Life

Species 2000

http://www.sp2000.org/

Federation" of database organisations working closely with users, taxonomists and

sponsoring agencies

GBIF

http://www.gbif.org/

Global Biodiversity Information Facility

IFREMER

http://www.ifremer.fr/anglais/

French Research Institute for Exploitation of the Sea

VLIZ

http://www.vliz.be/

Flanders Marine Institute

ICES

http://www.ices.dk/indexfla.asp

 Coordination and promotion of marine research in the North Atlantic.

Life Watch

http://www.lifewatch.eu

e-Science and Technology Infrastructure for biodiversity data and observatories

[ 21 ]

A2 Minimum Information for Neuroimaging Description and Specification (MINDS)

Several critical elements contributing towards MINDS include:

1 The raw data and full details for each MR acquisition type (e.g., T2, EPI, T1, DTI files; TR, TE, FOV, etc.).

2 The final processed (normalized) data for the set of acquisitions in the experiment/study (e.g., the version of

the data just prior to statistical analysis).

Full subject demographic and diagnostic details (e.g., age, gender, clinical group, etc.).

4 The experimental design including sample data relationships (e.g., functional time course design matrix,

conditional info, etc.).

5 Sufficient annotation of results with mappings to formalized neuroanatomical reference.

6 The essential data processing protocols and workflow (e.g., what normalization method has been used to

obtain the final processed data--provenance).

7 The details of how statistical modeling was performed and inferences have been made (e.g., GLM, GRF,

ICA, etc.).

[ ]

December 13, 2007

09.00 09.15

Introductions (Bjaalie and Van Pelt)

09.15 18.00

Scientific presentations and discussions

Jaap van Pelt

Introduction to the workshop Sustainability issues of neuroscience

databases

Shankar Subramaniam

Interoperability and Data Integration in Neuroscience Databases

Tadashi Isa

National brain research project "Integrating Brain Research" from the database

committee point of view

Shiro Usui

J-node sustainability scheme including government support

Jack Van Horn

Business Models for Neuroscience Database Sustainability

Chris Emblow

The Society for the Management of European Biodiversity Data and its role in

the sustainability of taxonomic checklist databases

Wouter Los

The Maintenance, sustainability and management of databases in the

environmental sciences of partial correspondences in a single species

Fiona Reddington

Multi-disciplinary Data Sharing: A UK Perspective

Jostein Kandal Sundet

Data exchange in international collaborations

Ilya Zaslavski

Long-term preservation of spatial information: research issues and supporting

infrastructure

Matias Palva

Dataformat and interface issues for sustaining multi-scale databases.

Martin Kersten

Scientific databases, the story behind the scene

Roman Moucek

EEG and ERP records - storage and processing

Martin Nawrot

Three challenges: data complexity, data format confusion and

non-transparency of data analyses

Alessandro Orro

Data management in the EGEE Grid Infrastructure: the BioinfoGRID

experience

16.30 17.30

Meetings of the Discussion Groups on "Community Issues", "Technical Issues"

and on "Standardization Issues"

19.00

Dinner

December 14, 2007

09.00 16.00

Discussions

10.00 12.00

Meetings of the Drafting Groups on "Recommendations" and on "Sustainability

Issues"