The fifth Joint Conference of the U.S. National Cancer Institute cancer Biomedical Informatics Grid® caBIG® and the U.K. National Cancer Research Institute (NCRI) Informatics Initiative took place in London at the Wellcome Trust on June 13-14, 2011.

This annual event is the result of a close international collaboration founded on the sharing of open-source tools, infrastructure, and data. This year, the meeting had a different format from those of previous years. Conscious of the fact that these conferences present a unique opportunity to formulate concrete collaborations (bringing together as they do, a wealth of expertise and knowledge from around the world) the conference moved away from the open registration format of the past, to being an 'invitation-only' event, with international representatives invited to contribute to one of five key themes.

Our hope was that the small size of the groups working on each theme would facilitate the quick and easy exchange of ideas between key international players, and, hopefully, the generation of actual collaborations which would help to address key, unresolved issues in the respective themes.
In this regard the conference was a success with each of the groups producing recommendations, commitments and collaborative proposals which will be taken forward in the coming months. These experiences will form the basis for a number of peer reviewed papers and, it is hoped, submissions to Nature Precedings.

The five themes (referred to as ‘tracks’) that formed the basis of this invitation-only meeting were:

• Sharing Data from Large-Scale Genomic Studies (Co-Chairs Adam Butler (UK), Paul Spellman (US))
• Imaging (Co-Chairs Darren Treanor (UK), Joel Saltz (US))
• Use of Clinical Data for Research Purposes (Co-Chairs Peter Knight (UK), John Speakman (US))
  
• Standards and Vocabularies (Co-Chairs Susanna-Assunta Sansone (UK), Stuart Turner (US))
• Tools and Technologies (Co-Chairs Jim Davies (UK), George Komatsoulis (US))

Opening Plenary – Progress Update

The conference opened with introductions to the group from Dr. Ken Buetow and Prof. Sir Alex Markham:

Dr. Buetow (Director of the NCI Center for Bioinformatics and Information Technology (NCI CBIIT)) spoke about the continuing need for integration between the various themes and disciplines within biomedical science to the end of improving clinical outcomes and empowering patients to manage their own health. He memorably adduced the example of the Midland Grand hotel, closed 59 years after opening as its eight bathrooms were insufficient for the changing requirements of the guests occupying its three hundred rooms, in asserting the necessity of building an infrastructure fit for the future.

	Biomedical Informatics: Past, Present and Future Directions.ppt

 

Prof. Sir Alex Markham (Chair of the NCRI Informatics Initiative, Chair of the National Cancer Intelligence Network (NCIN), and Chair of the Board for the NIHR Research Capability Programme) reflected on the achievements of these three organisations. He mentioned in particular the NCIN’s use of registry data to publicise the performance of NHS Trusts and drive service improvement; and discussed the issue of public involvement in and use of patient data in research towards personalised medicine as well as the IT challenges the latter presents, concluding that the issues which brought about the NCI/NCRI collaboration remain current today.

	NCRI Informatics caBIG® Presentation 2011.ppt

 

Overviews of the Breakout sessions

Sharing Data from Large-Scale Genomic Studies (Co-Chairs Adam Butler (UK), Paul Spellman (US))

 

Adam Butler
Adam Butler is currently a Principal Computer Biologist in the Cancer Genome Project (CGP) at the Wellcome Trust Sanger Institute.

Paul Spellman
A PI at Lawrence Berkeley Lab, Paul accepted a temporary assignment to work at the NCI on genomics issues, principally advising the Deputy Director on the matter of Cancer Genomics. In July 2011 Paul will move to Oregon Health and Sciences University where he will join the faculty of Molecular and Medical Genetics and become a member of the recently inaugurated Knight Cancer Center, Center for Spatial Systems Biology.

Paul Spellman gave the overview for this track indicating that the working group’s breakout sessions would address: legal and regulatory impediments to sharing data; cloud sharing/computing; the quality of data and structure annotation; and the development of a cancer cell line compendium database.

He exhorted the scientific community to exert itself in defense of data access against stifling legal frameworks whilst being mindful that patients need to be protected; and to look towards next-generation solutions to address the exponential increases in dataset scale. He also called for the creation of trusted ‘Data Brokers’ and for collaboration to prevent duplication of endeavour in relation to the development of cell-lines databases.

Paul presented the aims of this track with the following slides:

	Sharing Data from Large-Scale Genomic Studies.ppt

 

Adam summarised the outcomes of the two days’ sessions against the same subthemes with the addition of a fifth: tools.

In overview:

• The group concluded that simplification of access to data was required and called for the establishment of a body to certify researchers for access to managed-access data
• They established a level of commitment to the creation of Data Brokers for cancer sequence data
• The group proposed large scale multi-centre efforts to characterise cell lines to include genetic and functional data, a collaboration between the UK and US to prevent bifurcation

Adam presented the outcomes of this track with the following slides:

	Data Sharing.ppt

 

 

Imaging (Co-Chairs Darren Treanor (UK), Joel Saltz (US))

Dr. Darren Treanor

Dr. Darren Treanor is a consultant pathologist at Leeds Teaching Hospitals and honorary senior lecturer at the Leeds Institute of Molecular Medicine, University of Leeds. Dr. Treanor also runs the Leeds virtual pathology project and he is the principal investigator on the UK National Institute for Health Research (NIHR)-funded project entitled “A virtual reality microscope for pathology”.

Joel Saltz, M.D., Ph.D.

Dr. Joel Saltz is Director of the Center for Comprehensive Informatics, Professor of Pathology, Biostatistics and Bioinformatics, Mathematics and Computer Science at Emory University, Chief Medical Information Officer at Emory Healthcare, Adjunct Professor at Georgia Institute of Technology in the School of Computer Science and the Division of Computational Science, Georgia Research Alliance Eminent Scholar, and Georgia Cancer Coalition Distinguished Cancer Scholar. In addition, he serves as Director of the In Silico Center for Brain Tumour Research at Emory.

Dr. Saltz gave the overview for this track clarifying the scope (pathology imaging in clinical trials/ clinical research) and goals: to assess the importance of quantitative anatomic pathology; to characterise the role of imaging in the development of a quantitative pathology framework; to define the context and scope of the necessary research and to launch efforts to develop necessary methodologies. In doing so he asserted that developments in this field would most likely follow the pattern observed in Radiology Imaging Quantification: going from unstructured observations to reproducible measurement.

Dr’s. Saltz and Treanor had arranged their breakouts to cover image processing and validation; standards and image handling in whole slide imaging and TMA; integration with other sources of information including clinical, genomic, proteomic, radiology data; and finally credentialing, validation and provenance in clinical research use of pathology imaging.

Dr. Saltz presented the aims of the track with the following slides:

	Imaging (Pathology) Theme.ppt

 

Darren presented the summarised outcomes of the breakout sessions outlined above making a series of recommendations including:

• to provide public datasets to develop and quality control image analysis results
• to standardise “generic” annotation and markup conventions
• and to develop a minimal set of descriptors to support pathology diagnosis and integration of pathology information with the broader research/clinical context

The group had discussed useful tools for image management and harmonization and also defined possible approaches to integration with other sources of information including clinical, genomic, proteomic and radiology data.

With regard to credentialing, validation and provenance in clinical research use of pathology imaging, they gathered requirements to combat the variability of pathology imaging and looked at sources of this variability in clinical studies.

Darren presented the outcomes of this track with the following slides:

	Pathology Imaging.ppt

 

 

Use of Clinical Data for Research Purposes (Co-Chairs Peter Knight (UK), John Speakman (US))

Peter Knight

Peter Knight is Group Programme Director of NHS National Institute for Health Research’s Research Capability Programme.

 

Dr. John Speakman

John Speakman serves as Chief Program Officer of the Center for Biomedical Informatics and Information Technology (CBIIT) at the National Cancer Institute (NCI), within the United States Federal Government. John leads CBIIT’s programs, including the cancer Biomedical Informatics Grid (caBIG®) program, supporting molecular science, biobanking, in vivo imaging, clinical trials, population sciences and healthcare.

Peter Knight spoke on the aims of this track within which patient advocates, information governance professionals, pharma representatives and healthcare data managers from the US, England, Scotland, Wales and Australia discussed achievements and issues to be overcome in the area of data integration for research. The group agreed to work on a concrete collaborative project which would serve to highlight issues and potential solutions in this area.

Dr. John Speakman summarised the theme’s outcomes in the closing plenary session. He addressed the context within which research use of clinical data was undertaken:

• that the terminology “secondary uses” highlights the fact that data is not considered a multi-purpose core resource in healthcare in the way that it is in other industries
• that this makes clear the need to think about health data in a systematic way to the ends of improved public health and better outcomes
• and that privacy requirements and patient involvement need to be thought of in different ways to leverage the desire that many patients have to see their data put to use beyond their own immediate care

He also addressed:

• the technical and procedural requirements to enable data integration and analysis
• governance issues (and the key partnership between patients and users)
• the complexity of the standards landscape and technical architectures

The group ended by discussing use cases which led to a proposal which will be taken forward by various members of the group: looking at potential links between diabetes medication and the differential incidence of some cancers, they will integrate de-identified data from multiple locations [UK (England, Scotland, Wales), Australia, US] and from across disparate technology platforms.

John presented the outcomes of this track with the following slides:

	"Secondary Use".ppt

 

 

Standards and Vocabularies (Co-Chairs Susanna-Assunta Sansone (UK), Stuart Turner (US))

Susanna-Assunta Sansone Ph.D.

Susanna-Assunta Sansone PhD, is Team Leader at the University of Oxford e-Research Centre, Oxford, UK. She is co-founder of the Minimum Information for Biological and Biomedical Investigations (MIBBI) and the BioSharing initiatives, and sits on the board of several standardisation initiatives, including the Genomic Standards Consortium (GSC) and the Functional Genomics Data Society (FGED/MGED). She is also committee member of the European ELIXIR ‘Data Integration & Interoperability Committee’ and Industry Liaison Coordinator between the Open Biological Ontologies (OBO) and the industry-driven Pistoia Alliance initiative.

Dr. Stuart Turner

Dr. Stuart Turner is a Principal in an applied informatics consultancy (Leafpath Informatics, LLC).

Dr. Sansone gave the overview for this track which broke down into the following subthemes: biomedical resource representation and profiling; principles for adopting standards and vocabularies; interoperability scenarios; clinical models, serialisation and reasoning; and cross-realm harmonisation –  building alliances.

Dr. Sansone indicated that their sessions would cover a review of the activities of the various working groups active in this area; the relationship between data sharing policies and standards – particularly the Biosharing catalogue; social networks and collaboration; and representational artifacts. The intention was to work through use cases which reflected the expanding needs of researchers and to demonstrate how qualitative and quantitative metrics might be applied to standards; also to explore the influence of authorities and emerging policies on adoption. Interoperability scenarios were presented (e.g. BRIDG usage in real life) and used to highlight the issues thrown up by legacy data, compliance, emerging frameworks (SAIF) and the need to preserve the ‘context’ of discrete data sets.

A session was dedicated to clinical models, serialisation and reasoning in which tying meaning to data representation was discussed and the sessions concluded with cross-realm harmonisation: looking at the diversity of domains, their boundaries and the drivers for the development of standards.

Dr. Sansone presented the aims of the track with the following slides:

	Standards and Vocabularies.ppt

 

Work is still ongoing on this theme with attendees collaborating online in the creation of a set of outcomes and recommendations. Dr. Turner presented an overview of this work in progress at the final plenary session.

The group concluded:

• whilst ‘cross-community pollination’ was valuable and desirable, there are legal, technical and cultural barriers to this
• those working in standards and vocabularies need to acknowledge the plurality of the users they are targeting
• to make information more accessible and choices easier, profiling standards, ontologies and people is a key activity
• the group discussed VIVO for profiling people, Biosharing and BioPortal for standards and ontologies and the NCBO’s Ontology Recommender as an example of a web service which helps users to choose ontologies relevant for their domain

They have also begun to clearly document the many challenges which exist and should be recognised in order to improve adoption by the community. Their current document additionally contains a section on metrics and methods and their use in helping to evaluate the development of standards as well as to increase users’ confidence in their adoption.

The groups demonstrations included LOINC for the creation of survey instruments, HL7 for Clinical Genomics, BRIDG and LS-DAM; and the interoperability scenarios they worked through looked at imaging (deriving models and reconciling datasets) and at the OBO Foundry.

Further updates on this theme are expected in early July and will be posted here.
Dr. Turner presented this discussion with reference to the following material:

	Standards and Vocabularies.pdf

The following are further presentations from this track:

	HL7 Clinical Genomics - From Research to Healthcare.pdf
	ISA Software Suite: Supporting Standards-Compliant Experimental.pdf

Tools and Technologies (Co-Chairs Jim Davies (UK), George Komatsoulis (US))

Prof. Jim Davies

Jim Davies is Professor of Software Engineering at Oxford University. He serves on the Informatics Advisory Group for CR-UK's Stratified Medicines Programme and the Tools and Annotations Working Group for the Sage Commons. He is General Chair for the International Conference in Electronic Governance (ICEGOV) 2011.

George Komatsoulis Ph.D.

George Komatsoulis, Ph.D. is the Deputy Director of the NCI Center for Biomedical Informatics and Information Technology and Chief Information Officer (acting) of the National Cancer Institute.

Dr. Komatsoulis gave the overview for this track within which each of the following subthemes was accorded a breakout session: data acquisition (with an initial emphasis upon original or source data, rather than data transferred from other systems); data analysis (creating tools at an appropriate level of abstraction for the target users); data sharing; and data storage and security.

Within their data acquisition subtheme, the group set out to discuss experimental design, the simulation and review of tools at the design stage in research planning, the automatic configuration of acquisition tools from designs, standards for annotation, support for interconnection of data types and future interoperation.

The emphasis of their data analysis session was on ‘not reinventing the wheel’ and removing the need for duplicative home-grown technologies by providing tools and technologies at an appropriate level of abstraction for users.

The data sharing subtheme was set to cover tools and technologies for the location, transfer, and transformation of existing data as well as annotation and the role of metadata in transfer. They also aimed to look at patient recruitment and continuing engagement – consent and identity management.

Their final session on data storage and security (tools and technologies to facilitate secure access to datasets) was targeted at the future access to current platforms (for data requiring infinite storage), secure access to data and secure management of associations between data and patients (long-term development in the context of changing notions of privacy, consent, and rights of access) as well as digital rights management.

Dr. Komatsoulis summarised these subthemes under the overall themes of meeting current needs but ensuring future reuse, and reuse or extension of capabilities to the maximum possible extent. He presented this summary in reference to the following slides:

	Tools and Technologies.ppt

 

Prof. Jim Davies summarised the outcomes of the discussions held within these breakout sessions which made recommendations in respect to several activities including:

• the design of user interfaces
• the use of metadata in information systems
• applying standards to information systems
• social engineering via software (encouraging data sharing)

His summary made reference to the following slides:

	Tools and Technologies Recommendations.ppt

 

The following are further presentations from this track:

	RightField Semantic Spreadsheets for Science.ppt

	myGrid e-laboratories SysMO and MethodBox.ppt