Events

17 Oct 2014

View the Meeting Report

2nd Plenary Meeting

17 Oct 2014

GA4GH held its second major meeting on Saturday, October 18, 2014 in San Diego, California. The purpose of the plenary session was to convene the diverse community of GA4GH members and stakeholders, share important progress made by the four Working Groups and projects being accelerated, and collectively discuss next steps for GA4GH to advance and scale up current efforts in genomic data sharing to improve human health.

2nd Plenary

OVERVIEW

The Global Alliance for Genomics and Health (Global Alliance) held its second major meeting on  Saturday, October 18, 2014 in San Diego, California. The purpose of the plenary session was to  convene the diverse community of Alliance members and stakeholders, share important progress  made by the four Working Groups and projects being accelerated, and collectively discuss next  steps for the Alliance to advance and scale up current efforts in genomic data sharing to improve  human health. 

This summary document includes a brief description of key themes, followed by a more  comprehensive summary of each speaker’s remarks, featuring goals, accomplishments, next steps,  and participants’ questions and feedback. In addition, the Appendix of this document includes brief  descriptions of each Working Group’s all-day planning meeting and the cross-Working Group  meeting that took place in addition to the plenary session.

Key themes

There was strong consistency in the messages and key issues raised by the presenters, panelists,  and meeting participants. These included: 

  • Need for Rapid and Effective Action: The field is transforming rapidly, creating a vacuum that  needs to be filled. The Alliance is focused on making quick progress and addressing the biggest  challenges in this evolving environment. 
  • Progress on Key Tools and Flagship Projects: The Global Alliance developed key products  over the past six months that were needed in the field, and is poised to deliver even more in  2015. By launching flagship projects such as the BRCA Challenge and Matchmaker Exchange,  the Alliance is focused on demonstrating value and best practices. 
  • Overcoming Technical and Regulatory Barriers: Key tools now in place to tackle barriers to responsible data sharing include the GA4GH Genomics API and the Security Infrastructure.  Policies around responsible sharing and use will continue to be challenging, and are a key focus  of the Alliance, with the Framework serving as the policy foundation. 
  • Need to Incentive Use and Handle Misuse: The Global Alliance needs to actively promote and  incentivize use of methods to share data and will identify the best approaches to minimizing  misuse of data. 
  • Engaging Diverse Stakeholders: Diverse participation, both globally and sectorally, is vital for  success, and a central focus as the Alliance builds and engages membership. The Alliance must  also draw linkages not only to researchers and academics, but to the disease advocacy  community, private sector, and regulators, acting as a global convener and hub for all involved. 
  • Demonstrating Results: Now that the Alliance has developed a number of key deliverables, we  need to demonstrate the results of these actions and communicate about tools broadly.

WELCOME

Keith Yamamoto (University of California, San Francisco) opened the plenary meeting, framing  the work of the Alliance as remarkable and inspiring within the broader field of genomics and health  and at the “leading edge of precision medicine.” 

Yamamoto described the Alliance as holding the promise to help spark a revolution across research,  health, and healthcare. Yet, the gap between the data available and our understanding of its  immense complexity is vast. The challenge confronting the Alliance is how to move through that  information deluge, creating a “knowledge network” from the vast amount of data enabling precise  diagnosis and treatment decisions for each individual, empowering further research, advancing  clinical care, and informing patients and citizens. 

Yamamoto outlined three types of reinforcing actions to continue to drive the mission of the Alliance. 

First, the field requires significant additional data integration activities, including building the  technical ability to integrate, store, and analyze genomic information. 

Second, moving to a precision medicine approach will better serve the research and clinical  communities. Yamamoto proposed a thought experiment: what if disease was classified by its  genetic mechanism? One disease, such as breast cancer or diabetes, might be described by its multiple underpinning genomic mechanisms, just as one mechanism may be implicated in more than  one disease.  

Research and academic institutions in the Global Alliance network can advance this type of  discovery, promote the development of more accurate disease classification, and advance  diagnosis, therapy, and care. Operationalizing this research to health continuum would require four  elements: an information commons and knowledge network, embracing the goal of precision  medicine, building a layered knowledge network across disciplines, and drawing on network  connections to carry innovation and discovery to new areas of research and patients groups. 

Yamamoto described the power of a precision medicine approach in the case of Traumatic Brain  Injury (TBI) in the U.S., where genomics is becoming an increasingly important part of diagnosis and  treatment. Instead of the traditional one-size-fits all approach, some researchers are examining not  just imaging and clinical data, but patient proteome and genome data and discovering new linkages between TBI  expression and an individual’s genetic makeup.  

Third, Yamamoto described the need to foster system  integration not only between science disciplines, but all  along the biomedical continuum, and among all  stakeholders.  

Yamamoto stressed that the Alliance is foundational on all three of these planes, and is at the lead of being able to develop the idea of precision medicine into  practice. 

GLOBAL ALLIANCE PROGRESS AND GOALS

David Altshuler (Broad Institute of MIT and Harvard), Chair of the Global Alliance Steering  Committee, welcomed all attendees to the second plenary, and stressed the important moment for  the Alliance. Altshuler reminded attendees that the field is in the midst of a data revolution, and the  Alliance is addressing some of its greatest challenges: developing trusted routes for data sharing,  effective managing of privacy issues, and respecting the diversity of views for how to drive efforts  forward.  

Altshuler stressed that the stakes are high, and that it will be up to the people in the room and  involved in Alliance to make sure the field develops in the right way before we lose this window of  opportunity. But he stressed that there is great potential to unleash a new era of innovation in health. 

For over a year, the Global Alliance has been working to accelerate progress in human health by  helping to establish a common framework of harmonized approaches to enable effective and  responsible sharing of genomic and clinical data, and by catalyzing data sharing projects that drive  and demonstrate the value of data sharing.  

Altshuler discussed how the collaborative, iterative, and transparent model of the Alliance has a  history in the technology space, but is newer to healthcare. Nevertheless, the progress of the Alliance has been substantial. 

He noted that the Alliance itself has transitioned from a  nascent coalition to a formal membership organization,  with an established Constitution and composed of 141 exciting and unique organizational members in 23  countries to date. Altshuler also mentioned that diversity is  vital to the Alliance’s current and continued progress, and  one of the priorities for the coalition is to actively to  engage an even more diverse set of organizations and  individuals in the coming months.

Before wrapping his presentation, Altshuler looked to the future, and how a widespread willingness  to share data for the greater good is needed to realize the benefits of the work. He looked to the  cultural shifts and incentives that will be needed to realize greater sharing, while respecting privacy  and security of individuals.  

Altshuler closed by describing the Global Alliance as poised to deliver larger-scale products in the  coming months, driven by the community’s incredible passion and energy to advance human health.

REGULATORY AND ETHICS WORKING GROUP

Bartha Knoppers (McGill University) and Kazuto Kato (Osaka University), the Chair and Co Chair of the Regulatory and Ethics Working Group (REWG), introduced the Working Group’s  mission to address the ethical, legal, and social implications of enabling responsible genomic and  clinical data sharing. This includes preparing the overall policy Framework for data sharing, and developing forward-looking governance policies pursuant to the  Framework on consent, privacy and security, and accountability.  

Knoppers emphasized that progress in the past year has been substantial: the Regulatory and  Ethics Working Group launched six active Task Teams, fostered broad international engagement,  entered into discussions with multiple international consortia, authored numerous publications, and  created a major policy document, the Framework for Responsible Sharing of Genomic and Health-Related  Data. 

The Regulatory and Ethics Working Group Chair  described the six active Task Teams launched in 2014  that allow for the benefits of genomic research while  protecting individual privacy. 

  • Consent Task Team: developing core  elements of consent to enable responsible data  sharing, including clauses relating to privacy,  international research collaboration, and processes and methods for data storage. In  2014, the Team developed a draft set of Consent Tools in plain English and will develop  a Consent Policy in the coming months.
  • Data Protection Regulation Task Team: analyzing and rapidly responding to data  protection regulation development occurring around the world, as it affects biomedical  research and sharing of health-related data for research purposes. 
  • Data Safe Havens Task Team: developing policies for platforms for the secure and efficient  exchange of clinical and genomic data. The Task Team has developed consensus positions  for “Genomic and Clinical Data Sharing Policy Questions with Technology and Security  Implications”. 
  • Ethics Review Safe Harbor Task Team: developing an international system that allows for  mutual recognition of data access and ethics review. The Task Team is currently conducting a literature review and comparative analysis of ethics policies that will lead to an “ethics  review safe harbor” to allow for mutual recognition of data access and ethics standards.
  • Framework Task Team: successfully drafted and released the Framework for Responsible  Sharing of Genomic and Health-Related Data, containing foundational data sharing  principles and guidelines for research, guided by the human rights of all citizens to benefit  from advances in science and its applications, and the right for contributors to be recognized  therefore. The Framework has been translated into five languages to date (Arabic, Japanese, Chinese, French and Spanish) and is published Open Access in The HUGO Journal.
  • Privacy and Security Policy Task Team: developing a Privacy and Security Policy for  genomic and clinical data sharing in line with the Framework and core Global Alliance  policies and recommendations. 

Currently, the REWG is establishing policy-specific Task Teams to draft content-specific policies  guided by the Framework. This includes defining the key attributes of a data “safe haven” for secure  and trusted genomic and clinical data sharing, a review of ethics review regimes, and developing a  “points to consider” policy for establishing ethics review equivalency across intuitions. Knoppers  further announced that two new policy-specific Task Teams, one on Accountability and the other on Privacy and Security, are in the process of being formed, based on need and strong member  interest.  

Knoppers concluded her presentation by explaining that the REWG is poised to continue advancing  responsible data sharing in three avenues: 

  1. Supporting the creation of efficient and secure deposit, curation, and access mechanisms for  available datasets;  
  2. Establishing a system to respect the rights of participants, researchers, and others; and 3) Exploring the possibility from a regulatory and ethical perspective of cross-border data  linkage with health care providers and globally unique content-based identifiers for genomic  research and medicine.

Questions from the plenary attendees included which policies and international developments the  co-chairs are most closely watching. In response, the REWG Co-Chairs described concerns and  policy changes regarding the protection of personal data, not only in the European Union, but in the  United States, and around the world. Co-Chairs invited members of the Global Alliance to channel  information on personal data protection policies and news to the REWG to ensure real-time  engagement and response in this dynamic area. 

REWG next steps:

  • Developing three Policies: a Consent Policy, a Privacy and Security Policy, and an Accountability  Policy, all for release in 2015, to action the Framework
  • Ethics Review Safe Harbor Task Team conducting review of ethics policies

SECURITY WORKING GROUP

Paul Flicek (EMBL – European Bioinformatics Institute) and Dixie Baker (Genetic Alliance),  Co-Chairs of the Security Working Group (SWG), described the need to lay a foundation for securing  a data sharing commons, and to create a technology environment that provides individuals,  researchers, clinicians, and other stakeholders assurance that data made available are shared,  annotated, and interpreted only by appropriately authorized persons and entities in accordance with  Global Alliance security and privacy policies. The Co-Chairs described their goal to help create a  technology environment that provides such assurances, and the Global Alliance as a key enabler for  this type of productive security ecosystem to emerge.

The major product being developed by the SWG is the Global  Alliance Security Infrastructure, Version 1.0. The objectives of  the Security Infrastructure are to manage five types of genomic  data security risks: 1) unauthorized disclosure of stakeholder  data, 2) unauthorized access or use of stakeholder data, 3)  corruption or destruction of data, 4) disruption or degradation of  services supporting availability and access to data, and 5) that  inappropriate actions result in security incidents that diminish  participation in the Global Alliance ecosystem.

The Security Infrastructure provides recommendations for

security technology infrastructure to support Global Alliance privacy and security policies, Baker  announced that Version 1.0 of the document has been published on the Global Alliance website. The Infrastructure is a “living document,” and as such, has been published with a link to enable  submission of comments. 

Looking to the future, the Co-Chairs described four primary areas of focus for the SWG. The first is  establishing a collaborative effort for security incident reporting. Second, the SWG plans to develop  standards for consent management. Third, the Working Group looks to test the proposed Security  Infrastructure against existing and emerging sharing activities, particularly drawing on the expertise  and insights of other working groups. Finally, the group will provide cross-cutting advice to other  Alliance Working Groups and initiatives to ensure harmonized approaches to responsible data  sharing. Three new SWG Task Teams were also proposed: Incident Response, Software Security,  and Cloud Security.  

In response to a question about the amount and type of data that can be released before a privacy  breach, Flicek discussed reframing the concern from creating uniqueness from a data breach to  doing harm. Whereas creating uniqueness is a technical concern, the question of when a data  breach crosses over into harm is a policy one that should be further discussed. When asked a  question on incident handling, the SWG Co-Chairs described the rapidly changing area of  application security and, from a technical perspective, that such measures must be embedded in any  data sharing approach.

SWG next steps:

  • Establish three new Task Teams for Incident Response, Software Security, and Cloud Security • Solicit feedback on the Security Infrastructure, Version 1.0
  • Provide security support to other GA4GH Working Groups and projects

DATA WORKING GROUP

David Haussler (University of California, Santa Cruz) and Richard Durbin (Wellcome Trust  Sanger Institute) introduced the Data Working Group (DWG), describing the work of the group to  overcome siloed data in incompatible systems and break down barriers to collaboration. 

A global, accessible, and collaborative way of working defines the broader mode of operation of the  DWG, which utilizes an open source software development environment at https://github.com/ga4gh. All individuals are welcome to participate and decisions are made nimbly by those most active in the open source development process. 

In the past year, the DWG has launched five task teams to overcome key barriers to the technical  sharing of genomic data:

  • File Formats Task Team: custodians of the  SAM/BAM, CRAM and VCF/BCF file formats,  responsible for standards definition, evolution, and  reference implementations. A current goal of the Team  is to promote adoption of CRAM for data-sharing. The  CRAM format saves approximately 30% disk space  over BAM with lossless compression, but more  importantly enables much greater future savings  through managed data “smoothing”.
  • Genomic Reads Task Team: providing Application Programming Interfaces (APIs) to interoperably store, process, explore and share DNA sequence reads across multiple organizations and on  multiple platforms. Lead developers of the GA4GH version 0.5 API. Next steps include  releasing a reference implementation of the 0.5 API, developing browser integrations (3 are  currently underway), and developing batch processing integrations (currently for Spark and  Google Dataflow).
  • Reference Variation Task Team: developing and standardizing ways in which genomes are  represented and compared, and how genetic variants are reported. Basic variant reporting  relative to the current human genome reference is already featured in the version 0.5 API.  The longer-term goal is to develop with others a game-changing human genetic reference  that includes all common human genetic variation and provides a stable and consistent  method of representing both simple and complex genetic variation.
  • Metadata Task Team: focused on “everything but the sequence,” this Task Team is building representations and exchange formats for the non-genetic and non-clinical information that  comes along with genomes. The Task Team is preparing to turn to ontology work with the  Clinical Working Group, model refinement, and validation and data transfer tools.
  • Benchmarking Task Team: creating test suites and methods to measure the accuracy and  efficiency of genomics software. This Task Team has made dramatic progress since  launching in June. They are actively developing benchmarking tools that include a variant  comparison engine and working with leading groups to host data.

The Co-Chairs also announced the Working Group will incubate four new task teams, building on the  success of current efforts: 

  • RNA and Gene Expression Task Team: will provide APIs to interoperably store, process,  explore and share RNA sequence reads, computed transcript structures, and their  expression levels 
  • Genome Annotation Task Team: will develop standardized ways to represent information  associated with particular regions of the genome such as presence of functional units or  disease-associated variants 
  • Genotype2Phenotype Association Task Team: will formalize the language and methods  used to represent different kinds of genotype-phenotype associations and how confident we  are in our assessment of these associations 
  • Containers and Workflows Task Team: will concentrate on methods that increase the  portability of code and the reproducibility of computational analysis 

The Co-Chairs further described an exciting idea that was discussed at their group workshop the  preceding day: Globally Unique Content-based Identifiers or Digests. Haussler raised the idea of producing concrete technology so that any genome dataset in the world can have an abstract  identifier that is 1) unique, 2) privacy preserving, 3) not centrally assigned, 4) independent of the  computational representation of the data, and 5) unforgeably linked to the content of that dataset. This identifier system, based on a cryptographic hashing method, would be vital for verification, de duplication, and for auditable tracking of reproducible analysis and inference. The ideas extend to  other data types beyond genome sequence data.  

When asked about the DWG’s engagement with research funders, the Co-Chairs responded that the  group is engaging additional funders in several countries who are eager to participate in the Group’s  activities. David Altshuler added that the Alliance Secretariat is working to secure funding to support these advancements, and asked the Alliance community to put the group in touch with additional  funding agencies. 

When Alliance members at the plenary raised that the rare disease community is also considering  the development of globally unique identifiers, the distinction between identifiers for individuals and  identifiers for datasets was discussed (the digest concept is the latter) and it was proposed to review  the cross-working group implications of these efforts and report out on this issue in the coming  months. 

In response to a final question of how available the tools of the DWG are to the broader global  community, DWG Co-Chairs explained that these tools are available to anyone who can access the  Internet, but acknowledged that development of tools must be combined with education and  outreach to increase contributions and uptake. Once the reference implementation is complete,  Haussler stated that the DWG will be in a stronger position to reach out actively and broadly to  potential users.

DWG next steps:

  • An open source reference implementation for the Genomics API in 2015
  • Tools including a benchmarking suite and variant comparison engine in 2015 • Fully launching four new Task Teams: RNA and Gene Expression, Genome Annotation,  Genotype2Phenotype Association, and Containers and Workflows
  • Exploring the concept of dataset Digests with the other Working Groups and outside organizations

CLINICAL WORKING GROUP

Kathryn North (Murdoch Childrens Research Institute) began by noting that the Clinical Working  Group (CWG) is driven by one question: how do we represent phenotypic data and link it to  genotypic information? The goal of the CWG is to address both the research and clinical use of  genomic data, utilizing an approach that is physician-oriented, researcher-focused, and patient centered. 

North described the major activities of the CWG, including the current mapping of initiatives that  promote data sharing. In addition to working closely with the BRCA Challenge and the Matchmaker  Exchange, the CWG hosts the following four task teams. 

  • Catalogue of Activities Task Team: mapping current initiatives that promote or enable data  sharing, starting with Mendelian Genetic Disorders and turning next to cancer, because of  the immediate therapeutic implications, and an eHealth catalogue of activities. 
  • Genomic Data and Electronic Health Records Task Team: working to better integrate  genomic medicine and research into  electronic health records. The CWG is  currently mapping existing and emerging  initiatives with the Global Genomic Medicine  Collaboration (G2MC) and others.
  • Clinical Cancer Genome Task Team:  working to harmonize the many sequencing  efforts in the cancer community by promoting standards and best practices that support  clinical decision-making. Near-term goals of  this Task Team include working closely with  existing initiatives.
  • Phenotype Ontology Task Team: bringing together international efforts to develop and  promote standard language and tools for recording patient clinical phenotypes for  diagnostics and translational research. 

In response to a question about the next steps for linking the CWG’s cutting-edge genomics work to  electronic health records, North responded a likely path would be selecting a few demonstration  projects, sparking a lively discussion about the need to involve major private electronic health records providers (several of which had already been invited to participate). Alliance leadership  agreed that while the CWG will remain technology agnostic, the participation of the private sector is  of critical importance. 

A second area of discussion centered on a question of integrating non-human model organisms into  the work of the CWG. North agreed to explore this important idea, which was also raised in the  context of the DWG’s Metadata Task Team.  

Plenary attendees also raised questions about better integrating disease into the CWG’s approach.  In response to a question about the place of infectious disease in the CWG, the Alliance agreed to  convene a group to explore the idea of an Infectious Disease Working Group or Task Team in the  next months. Those interested in exploring this idea were invited to volunteer or suggest others who  should be involved.  

Finally, in response to a question about how the CWG plans to engage the patient community into  the future, North looked to explore ways to encourage the patient community to be more active in  responsible sequencing and data sharing.

CWG next steps:

  • Catalogue of Activities Task Team is developing a catalogue of eHealth activities • Expanding phenotype ontologies 
  • Developing Demonstration Projects to incorporate genomic data into Electronic Medical Records across multiple sites 
  • Exploring integrating infectious disease, the private sector, the patient community, and non human model organisms into activities

BRCA CHALLENGE

Sir John Burn (Newcastle University) and Stephen Chanock (National Cancer Institute), BRCA  Challenge Steering Committee Co-Chairs, introduced the newly launched BRCA Challenge, the  mission of which is to translate the rapid expansion of sequencing capacity into useful knowledge  and, in particular, learn how to rapidly interpret variant data to generate clinical utility.  

The Steering Committee Co-Chairs described the BRCA Challenge as a vanguard effort to  aggregate BRCA1 and BRCA2 data in order to understand variation and its impact on human health,  while demonstrating how to approach large datasets for other disease areas of study. Since its  recent formation, the BRCA Challenge has formed a Steering Committee and is formalizing strategic  goals, a structure, key deliverables, and an aggressive timeline for progress.  

The BRCA Challenge announced its first steps will be  ensuring that three major datasets (ClinVar, LOVD, and  UMD) are queriable. The group then plans to expand to  include other datasets (like ENIGMA, CIMBA), and seek  out additional sources of data to add to the Challenge  as well. 

Burn and Chanock proposed releasing the three deliverables:  

  1. Population based assessment of allele frequencies of variants using available sequencing  resources; 
  2. Federated data collection of Pathogenic Variants for BRCA1/BRCA2, both building a structure for data sharing and establishing a common terminology; and 
  3. (In the long term) Improve and refine information for use in clinical interpretation of mutations 

The Co-Chairs concluded that advancing the BRCA Challenge, an effort that has captured the  imagination of many in the room and around the world, in the longer term requires sustaining  infrastructure, retaining expert leadership, and providing opportunity for grants and contracts to  advance knowledge. 

BRCA Challenge next steps:

  • Steering Committee laying out goals, deliverables, and a timeline for advancement • Ensure three major datasets are able to be queried in 2015, with further expansion

MATCHMAKER EXCHANGE

Heidi Rehm (Harvard Medical School) discussed Matchmaker Exchange, one of the key projects  that the Global Alliance is working to accelerate. Matchmaker Exchange is already working to bring  together cases with overlapping phenotype and candidate genes both to gain research insights to  return information to patients.

Rehm described how conceptually, matchmaking occurs when a submitter queries all API-linked  databases with gene candidates, disease name, and the submitter’s. If a match occurs, the  depositor and the requestor are both notified and details of case are shared manually to begin  follow-up studies to validate the match.

Rehm announced two major developments with  the project. First, the development of API Version  1.0, which is already forming an interface  between several distinct genomic databases.  Second, Rehm announced the creation and  launch of a new website www.matchmakerexchange.org which will form the basis upon which the group communicates  with the broader community of interest.

She noted that these accomplishments are being accelerated through close collaboration with each  of the Alliance Working Groups: the DWG providing support for the API development, the REWG  providing feedback on consent protocols, the SWG providing guidance on query authentication, and  the CWG sharing expertise with phenotyping for matchmaking. 

To build on this initial progress, the Matchmaker Exchange project plans to finalize the API in  conjunction with the GA4GH Data Working Group and to develop guidance for groups without a  database wishing to choose a site for data deposition and matchmaking support. 

Matchmaker next steps:

  • Finalize API for database queries
  • Develop guidance for groups without a database seeking matchmaking support

BEACON PROJECT

Steve Sherry (National Center for Biotechnology Information) presented the accomplishments of  the Beacon Project, a project first proposed at the Alliance’s March 2014 plenary meeting as a way  to demonstrate the leadership, institutional permission, and technical ability to share data. Beacons  also show where data is present, and where gaps in mapping may need to be addressed.  

Sherry noted that since the initial proposal in March 2014, fifteen beacons are now active around the  world, with more to come. The beacons themselves are an API query that indicates if a simple allele  is present in an affiliated institution’s data repository, and the group announced it will soon release a Beacon of Beacons, in which a single beacon query searches each database in a federation of  repositories. Sherry issued a call to the Alliance community to join in the project to link additional  dozens and hundreds of data repositories to the effort. The Beacon Project is also collaborating with  data repositories to finely tune beacon queries to adhere to individual repository concerns about data  sensitivity or consent requirements.

Overall, the Beacon Project has established three  tiers of beacon queries, based on the strength of  an affiliation between users and data repositories.  The most basic relationship is with an anonymous  user who can gain access to genomic information  that can be provided without privacy concerns.  Sherry mentioned that users who provide a verifiable name and institutional affiliation may  access additional genomic information that may  be scientifically useful, and in the final case, a

user may enter into a binding agreement for access to one or more specific datasets, including the  commitment to abide by clear and robust privacy standards, and gain full access to the relevant data.  

Plenary attendees raised several questions about protections for data and the possibility of donor  identifiability in the case of forensic or law enforcement queries and other scenarios. Currently,  beacons only point users to aggregated information from a group of studies, and not to individual  level data. However, Alliance leaders agreed that even providing aggregate information is delicate,  requiring guidelines for what types of aggregated data can be released, its implications, and the risks  of identification.  

Beacon next steps:

  • Project team will finalize and release a Beacon of Beacons in the near future
  • Global Alliance will convene a group to further develop guidelines for data donor privacy protection

PANEL ON ENABLERS AND CHALLENGES TO DATA SHARING

The final session during the plenary meeting was a panel discussion moderated by John Mattison (Kaiser Permanente) and featuring panelists David Glazer (Google), Brad Margus (A-T  Children’s Project), Johan den Dunnen (Leiden University), Cindy Bell (Genome Canada), and  Lana Skirboll (Sanofi). These diverse panelists were selected because of their different vantage  points throughout the world of genomics and their alternative perspectives on big, new opportunities  to advance the Global Alliance.  

John Mattison, as moderator, led off the discussion noting that while genomic complexity is vast,  putting datasets together as the previous presentations described allows us to understand incredibly  complex interactions in a more meaningful way. Mattison noted that privacy concerns are still of  critical importance to this field, but over time, social norms may evolve and shift the value proposition  in favor of increased sharing. 

Lana Skirboll, who spent her career at NIH before entering the private sector at Sanofi, raised a  number of key issues for the field. She described industry as being in the middle of a transition, increasingly recognizing the importance of data sharing and open innovation through increased  engagement with public-private partnerships, working with academics, and others. Skirboll  mentioned that governments are also transitioning, as they begin to realize that data are an  economic asset, restricting movement across borders, and that these are political developments to  pay attention to and engage in the context of trade treaties and other means. 

Skirboll also raised the issue of law enforcement access to data, mentioned in an earlier  presentation, and believes that access is inevitable and suggested this community focus on  identifying the risks and defining informed consent policies. Another emerging area of data  protection is for the data emerging from the exploding field of real-time, personal sensors. Finally, in  this shifting world, Skirboll recommended that the Global Alliance consider working more closely with  regulators to ensure approaches are informed and aligned. 

David Glazer described what he called a “thought-experiment” to illustrate key areas of focus for the  Global Alliance. Glazer asked: If the genomics and health community wanted to put together a database of 100,000 genomes that was useful and available in a matter of months, what he called a  Tree of Life for researchers, what would be required?  

Glazer noted that the first area to tackle would be the technical sharing of data, and this would likely  require a federated project to connect data repositories and quickly get to scale. Data hosts would  form the branches of the structure, filled in with sets of data provided by many data collectors and  contributors. Contributors could include highly mobilized individuals and groups, like Autism Speaks  or Brad Margus and the A-T Children’s Project, whose energy and resources would be vital. But  Glazer emphasized that this is already technically doable. 

What is more challenging, Glazer argued, is the policy, access, and consent rules that govern the  sharing and use of this information. Glazer concluded that in his opinion, the best way forward on the  policy side is to develop a portable consent process where data donors can choose to make their  data available to all qualified researchers, without the researchers needing to apply for access to  every “branch of the tree” individually. 

Brad Margus shared his family’s own struggle with disease, as two of Margus’s young sons were  diagnosed with ataxia-telangiectasia (A-T) a rare and devastatingly degenerative disease. Margus has since founded the A-T Children’s Project and devoted himself to coordinating and seeking funds for research on A-T. Recent advances in data sharing helped Margus to identify individuals whose  genomic information may hold a key for understanding more about this disease. Furthermore,  Margus has begun raising funds to perform additional sequencing, another avenue that has only  been available in recent years. From the perspective of a leading disease advocate, Margus  reiterated the importance of standardized, flexible consent forms to aid in the responsible collection  and use of genomic data to fight disease.  

Cindy Bell described her perspective as a funder in the field of genomics and health. Echoing Keith  Yamamoto’s opening remarks, Bell described the genomics and health community as being at an  inflection point, where additional impact must be demonstrated. Bell praised the incredible energy of  Global Alliance contributors, recognizing the voluntary efforts that have driven much of the coalition’s  progress to date.  

Yet Bell acknowledged that volunteerism, while a powerful demonstration of commitment, is not  sustainable in the long run and that targeted investments are vital. Bell described her organization’s own launching of a pilot project to support the objectives of the Global Alliance in Canada and  encouraged similar investment. Remarking on the shifting field, Bell concluded that costs have  shifted in recent years from sequencing to data management and infrastructure, and noted that  funders are increasingly supportive of public-private partnerships to achieve impact. 

Johan den Dunnen, the final panelist to offer remarks, spoke next from the perspective of database  management, as the founder of the Leiden Open Variation Database. He stressed that DNA  diagnostics, DNA knowledge, is based on sharing information on genes, variants and phenotypes,  and emphasized sharing. Years ago, surprised and frustrated by the lack of data sharing, den  Dunnen embarked on a remarkable personal project to create what is now LOVD, an open source  database of DNA variants. Den Dunnen described a culture that is shifting, but yet still largely hasn’t  overcome resistance to uploading data or attracting funding for this essential work. He suggested making sharing data obligatory by law with associated payment to host and curate the data. 

After these brief panel remarks, a question from the audience about how the Global Alliance might  successfully handle instances of data misuse and regulate the field sparked a lively discussion about  responsibilities and priorities. Alliance Steering Committee Members, including Altshuler and  Knoppers, agreed this is an issue of great significance for the future of the Alliance, and agreed that  this would be a future focus of the effort to discuss how best to handle these issues as a group.

CLOSING REMARKS

Martin Bobrow (University of Cambridge, Emeritus) delivered closing remarks at the plenary  meeting, summarizing the productivity of the coalition and energetic mood of the meeting. Bobrow described the Alliance as both a philosophical discussion and social movement for responsible data  sharing, connected to an even broader network of researchers, clinicians, patient’s advocates, and  more.  

Ultimately, Bobrow said, the Alliance will succeed or fail not on its philosophy, but rather on its ability  to produce transformative ideas and products that are both high quality and highly relevant. To have  this impact, the challenge for the Alliance is choosing the most impactful areas of genomics and data sharing to focus on in the coming months, amidst so many promising avenues. Many new and  exciting ideas were shared at the plenary meeting today, and members of the Alliance will be taking  up a number of them.  

Bobrow pointed out that one of the successes of the Global Alliance is that so much has been achieved to date through an essentially voluntary effort, proof that this work is highly important and  interesting to this field. Although, to have transformative impact over time, this work must be  sustained by funders.  

Bobrow concluded by thanking the attendees, participants, and Alliance leadership, and invited all to  the next plenary meeting in 2015.

APPENDIX: WORKING GROUP AND CROSS CUTTING PROJECTS MEETING SUMMARIES

Clinical Working Group • Sunday, October 19, 2014

The Clinical Working Group continues to evolve. The Catalogue of Activities for Mendelian Genetic  Disorders represents a major deliverable in 2014, on which several subsequent Work Products will  be modelled. The following Task Teams reported on their work and next steps: 

Phenotype Ontology Task Team 

  • Will assemble a multidisciplinary group to promote and advance the creation of phenotype  ontologies in the areas of cancer and common disease. 

Clinical Cancer Genome Task Team 

  • Will convene a group of leaders involved in known cancer data sharing pilots and efforts in  order to share best practices and assess whether there is opportunity to collaborate. • Will create a list of activities (Catalogue of Activities) for patient reported outcomes and  crowd sourcing. 

eHealth Task Team 

Many tasks were identified and assigned to move forward including:

  • Define and prioritize various use cases;
  • Define challenges and opportunities for analytics/interpretations when operating across data types;
  • Create a Catalogue of Activities specifically for eHealth;
  • Provide guidance for low concordance of exomes;
  • Identify issues for pedigree data collection;
  • Establish open source / crowd-sourced commons;
  • Identify data storage issues;
  • Establish methods to identify conflicting interpretations;
  • Identify best practices for family history data collection; and
  • Summarize and track the evolution of patent policy issues.

Future Goals and Points of Agreement 

The CWG is looking to engage the broader interested community and solicit advice and feedback in  San Diego. A few recommendations raised in San Diego include: creating a website similar to the  DWG Github to showcase current work and more easily allow others to participate, formalizing a  communication strategy (e.g. regular newsletters, greater engagement with other parts of the world  such, etc.), formalizing cross communication among Working Groups, convening meetings in other  parts of the world, and generating greater engagement with clinicians and scientist. These comments reflected both an awareness on behalf of the CWG as well as the desire on behalf of the  interested community to see that the CWG is more transparent and more accessible. 

Next Steps 

Potential next steps that were discussed included: 

  • Creating a somatic equivalent to the BRCA Challenge (i.e. BRAF); 
  • Defining clinically actionable genomes; and, 
  • Returning incidental germline findings.  

Before deciding next steps in these areas, the CWG will investigate what efforts currently exist and  then determine whether to create a new deliverable/project/task team. 

Data Working Group • Friday, October 17, 2014

In 2014, the Data Working Group released the GA4GH API version 0.5 and the API continues to  improve and evolve. There is broad enthusiasm for moving both existing and new the initiatives  forward. The following Task and Project Teams reported on their work and next steps: 

Reads and Reference Variation Task Team 

  • A graph data model is a work-in-progress; a v0.1 graph API using this data model to be  defined. 

Benchmarking Task Team 

  • Developing tools to benchmark performance of human SNP and small indel calls. • Developed a document describing definitions of SNP/indel performance metrics and how to  stratify performance by variant type, sequence context, and functional region. • To develop a variant comparison engine over the next 3 months, which will have the ability to  do vcf-vcf and vcf-database comparisons. 

File Formats Task Team 

  • Official GA4GH adopted file formats: BAM/CRAM/VCF/BCF. 
  • Moving forward, promote adoption of CRAM. 

Metadata Task Team 

  • Progress moving forward and aligning several data elements as well as API development. • Next step will be to apply the metadata specifications to driving biological use cases (e.g.,  BRCA, or cancer in general). 

Beacon Project 

  • New NCBI Beacon service offers responses with Public, Controlled and Full levels of  access. Properties of answer may depend on relationship between user and repository. • Next steps: further develop the API and attract new beacons. 

RNA and Gene Expression Task Team (NEW) 

  • provides APIs to interoperably store, process, explore and share RNA sequence reads,  computed transcript structures, and their expression levels  

Genome Annotation Task Team (NEW) 

  • Develops standardized ways to represent information associated with particular regions of  the genome such as presence of functional units (coding exons, regulatory elements, etc.) or  disease-associated variants.  

Genotype2Phenotype Association Task Team (NEW) 

  • Formalizes the language and methods we use to represent different kinds of genotype phenotype association and how confident we are in our assessment of these associations. 

Containers and Workflows Task Team (NEW) 

  • Concentrates on methods that increase the portability of code and the reproducibility of  computational analyses 

Future Goals and Points of Agreement 

The concept of globally unique content-based identifiers or digests was defined. Any version of any  genome sequence dataset (or other large dataset) in the world can have an abstract identifier that is: 1. unique for that dataset version (no “copy” of that sequence+metadata at any other location at  any time in the future will ever have a different identifier, and no two different versions will  ever “collide” by accidently getting the same identifier) 

  1. content-dependent (the same, i.e. semantically identical, genetic sequence content stored in  a different format will have the same content-based identifier) 
  2. privacy preserving (given only the identifier you can’t determine anything significant about  the genome the data are derived from unless you are allowed to use that identifier to retrieve  the genome itself) 
  3. not assigned by any central authority (each medical center can generate their own ids) 5. unforgeable (if I send you Jill’s genome with Sarah’s id, you will know immediately that  something is wrong) 

Next Steps 

  • Consensus on the addition of four new Task Teams (listed above) in addition to the existing  five Task Teams. 
  • Overlap with key projects is likely to increase. 

Regulatory and Ethics Working Group • Sunday, October 19, 2014

In 2014, the Regulatory and Ethics Working Group published the Framework for Responsible  Sharing of Genomic and Health-Related Data. It is referenced in the Global Alliance Constitution.  The Working Group also published Consent Tools (see below). 

Framework Task Team 

  • The Framework has been translated into 5 languages to date (Chinese, Japanese, French,  Spanish, and Arabic) and is currently being translated into Greek and German. It is available  open access on the Global Alliance website and in The HUGO Journal. 

Data Safe Havens Task Team 

  • Developed consensus position statements for “Genomic and Clinical Data Sharing Policy  Questions with Technology and Security Implications.”  
  • Findings have been relayed to the Consent Task Team and the Privacy and Security Policy  Task Team. 

Consent Task Team 

  • In concert with P3G-IPAC, three Consent Tools have been prepared for international data  sharing (legacy consent; consent clauses; consent template); all require adaptation  according to local social, cultural and legal specifications. 
  • Currently developing a Consent Policy in line with the Policy template in Appendix 2 of the  Framework. 

Ethics Safe Harbor Task Team 

  • Developing an “ethics review safe harbor” that would allow for mutual recognition of data  access and ethics review. 

Data Protection Regulation Task Team 

  • Keeping track and responding to global data protection regulation developments and actively  working with the Beacon Project group on sensitive data issues and a privacy test. 

Privacy and Security Policy Task Team 

  • Developing and Privacy and Security Policy that operationalizes the Foundational Principles  and Core Elements of the Framework by offering tools, benchmarks and best practices to  guide and inform responsible data sharing and governance processes as it relates to data  privacy and security. 

Future Goals and Points of Agreement 

The Consent Task Team is also currently discussing with GSK representations/Harvard MRCT  participants how the Consent Tools can inform consents for clinical trials and industry standard  consent forms. Additionally, the Consent Task Team is looking at the concept of machine-readable  consents (noting that HL7 is working on machine readable consent directive, as is the World  Economic Forum).  

In the future, the Data Protection Regulation Task Team plans to explore genomic cloud computing  and data protection issues.  

Noting that a paper has been written by REWG Executive Committee member Paul Burton and  colleagues (currently undergoing peer review) that explores the topic of “data safe havens”, it was  agreed that a “data safe haven” is a place where data can be stored and accessed by all types of  groups, that can be trusted by all parties, and that hosts genomic and clinical data, in both open and  controlled formats. 

On the subject of identifiers, there is some confusion about what the various types of identifiers are  in concept and practice (e.g. UUID, GUID, ORCID), including from a regulatory and ethical  perspective. Therefore, the REWG supports the idea of working with the other Working Groups to  develop a glossary or short document that explains what each of these terms means in concept and  practice, and how they may impact on the work of the Global Alliance. Not only would this document  better frame the discussions currently being had within projects, Working Groups, Task Teams, and Member Organizations, but it would also aid the genomic and clinical data sharing  community more broadly. Indeed, as identifiers (especially universal identifiers) are usually assigned by some agency or consortium with governmental or international oversight, this may be a space  where the Alliance wants to set leading standards for identifiers for genomic and health-related data.  To do so, it is necessary to speak the same language and understand what these various terms  mean. 

Next Steps 

Future ideas for possible development include: 

  • Exploration of the categories for data access (i.e. public/open, registered/user-identified,  controlled/user signed agreement) and unique identifiers (e.g. UUIDs, GUIDs). 
  • Creating a new “Accountability Policy Task Team” in late 2014 that will develop an  Accountability Policy, in line with the Policy template in Appendix 2 of the Framework. The  Policy will explore the characteristics of accountability within the Global Alliance (e.g.,  identifying relevant stakeholders, framing violations of various types as properly being  handled by organizations, and speaking about consequences of data misuse through  community norms). Tiers of accountability as well as scenarios and appropriate responses to  data misuses may be explored and addressed in the Policy. 

Security Working Group • Sunday, October 19, 2014

The major product of the group has been the Security Infrastructure document (to be released as  v1.0, with an open link for comments). Other Task Teams are currently developing a need for SWG  input. 

The Working Group decided to establish three new Task Teams: 

  • Incident Response Task Team (NEW)
  • Software Security Task Team (NEW)
  • Cloud Security (NEW)

Future Goals and Points of Agreement 

The Global Alliance currently cannot accept responsibility for security oversight (corresponding to  Control Objectives 4 and 5 in the Security Infrastructure). Two proposed models were proposed to  address this: 

  • Financial Services Information Sharing and Analysis Center (FS-ISAC, https://www.fsisac.com/) – a member-owned, non-profit entity that serves as the global  financial industry’s go-to resource for cyber and physical threat intelligence analysis and  recommendations, and anonymous information sharing
  • Security Incident Response Trust Framework for Federated Identity (Sir-T-Fi,  https://www.terena.org/mail-archives/sirtfi/pdfMAM3MjGx3A.pdf) — a draft document  developed by Security for Collaborating Infrastructures group of the Interoperable Global  Trust Federation (IGTF) that specifies requirements and metadata for managing security  threats and incident response in the absence of overarching governance or uniform policy
  • Other possibilities mentioned were the College of American Pathologists (CAP) lab certification program, and the International Standards Organization (ISO) 27000 series. 

Response to and mitigation of security breaches needs to be developed, including:

  • Controlled message template response 
  • Begin defining use cases for realistic breaches 

A technology solution for consent management also needs to be developed. Toward this objective, Eve Maler, Chair of the Katara User Managed Access (UMA) Work Group, presented ongoing work  to define a profile of the OAuth 2.0 authorization standard (IETF RFC 6749) that enables individuals  to authorize access to resources that they own.  

Going forward, the SWG will launch three new Task Teams focusing on Software Security, Incident  Response, and Cloud Security. In addition, interaction with the Task Teams of the other GA4GH  Working Groups will increase. The interaction will be achieved in several ways: 

  • Embedding at least 2 SWG members within Task Teams requiring SWG support. • Inviting Task team members to the bi-weekly SWG calls. 
  • Participating in other GA4GH Working Groups. Very Large projects to be tackled by SWG as  whole if necessary. 

Next Steps 

In the near future, form the three new Task Teams, drawing from the SWG Interest Group as  appropriate. 

Cross-Cutting Projects Session • Sunday, October 19, 2014

The four current Working Groups are catalyzing key collaborative projects that aim to share real world data. The Project Teams move their work forward autonomously, with varying levels of  coordination support and oversight from the Global Alliance, and drawing on expertise from the  Working Groups as required. It is a considerable achievement that these projects have been  initialized and are moving forward through uncharted territory as we continue to define relationships  and support mechanisms between key projects and the Global Alliance. 

Matchmaker Exchange  

  • Uses a single API to interface with many similar databases 
  • Guidance is requested from REWG on proposed informed consent for different tiers  (“Levels”) of matchmaking: no consent for Gene + high-level phenotype; consent required for  sharing variants, genomic files and detailed phenotype.  
  • Guidance is requested from SWG on query authentication and verification steps, security  requirements on the exchange.  
  • From CWG, need to engage expertise on phenotyping for matchmaking, will need improved  specificity and scoring for phenotype queries as datasets enlarge. A potential quick win for  cancer communities would be to do similar for somatic mutations, which are open source,  less concern about re-identification. 

Beacon Project 

  • Started as a simple test of the willingness of international sites to share data with a “yes” or  “no” answer to querying whether the “beacon” has variant X.  
  • “Beacon of Beacons” is being developed (DWG); it aggregates beacons and answers the  question does any Beacon have genomes with a given allele at this position.  • There are a growing number of beacons and synergy between Matchmaker Exchange and  Beacon Project and potential for greater alignment should be explored. 
  • Input requested from REWG and SWG on proposed data access levels and the option to  register and identify users (identity credentials/identity proofing); aspiration is a trusted  repository. 

BRCA Challenge 

  • Aims to translate the rapid expansion of sequencing capacity into useful knowledge and  learn how to rapidly interpret variant data to generate clinical utility.  
  • Global outreach and engagement will be important.  
  • API and ELSI issues to be developed and addressed by DWG and REWG, respectively.

Next steps

Working Group Coordinators will play a connecting role in ensuring that key project needs are  addressed in a timely manner.

Categories

Latest Events

16 Sep 2024
12th Plenary
Plenary
See more
21 Apr 2024
GA4GH Ascona Connect
Connect
See more
19 Sep 2023
11th Plenary
Plenary
See more