-
Notifications
You must be signed in to change notification settings - Fork 4
Minutes_2020_02_06
Utrecht Vredenburg 6th February 2020
-Presentation results so far of each WP -what has been done -what still needs to be done
-Integration WP2/3/4 -Demonstrator projects -Connecting to daily practice (VKGL, VKGN, KMBP……) -Next steps for guidelines
FAIR genome spreadsheet stakeholder needs
Attendance: nienke vd stoep, terry vrijenhoek ,janneke Weiss, jasmin bohmer, saskia hilteman, david van zessen, daoud sie, hinri kerstens, patrick Kemmeren, dieuwke roelofs-prins, lennart Johansson, daphne Stemkens, susanne rebers, gurnoor singh, joeri vd velde, jeroen Belien, esther van enckevort; marielle van gijn
Minutes by: Jasmin Boehmer, UMCU
-
welcome
-
intro round of the meeting members -> a few new and external participants that want to be involved -> Marielle stepped up in organising and chairing the issue since Morris is on sick-leave
-
FAIR genome mailing list should be updated with missing and new members, Marielle will take care of that
UP UNTIL NOW: Ppt slides on Github
- the “Draft of Recommendations for WGS in genome diagnostics for rare disease” should be part of the FAIR genome project, suggested by Jeroen
- Patrick asks why it’s limited to rare disease; explanation: rare disease projects are always used as super-projects as lead examples on european level
- Patrick suggests that the GA4GH guidelines are very america driven, however the tools are fine to apply for our purpose as well
- iCRF generator of Health-RI is unifying standards and it’s worth considering that all FAIR genome codebooks are published at Nictiz
- 1+MG project is worth collaborating and integrating with in opinion of Jeroen
- add link from that shared google doc Jeroens refers to on slide 1
NEXT STEPS:
- guideline development of all streams and reports of technical implementations in the local project partners’ institution
- inventorisation of pipeline adaptations necessary, Jeroen suggest a detailed documentations
ADDITIONAL:
- 1+MG project receives no direct EU funding, each member state has to chip in
- Jeroen has been asked to co-lead WG3 of the 1+MG for “Common standards and minimum dataset for clinical and phenotypic data”
- every participating country has to set-up a national mirror group
- FAIRgenome project will be used for NL input for 1+MG project
- consensus is that FAIR genome & 1+MG should align to minimise double efforts in communication, outreach, and input
- NL action plan is parallel to the 10 1+MG work package leaders
- the 1+MG roadmap will feed into the Europeans Commisions’ Horizon Europe funding programme that start 2021
- Europeans reference networks are known to the FAIR genome project members, and Marielle is part of the rare disorders reference network (clinical not research)
- the european data protection officer has to approve the ERN’s
- Jeroen suggests to create a mindmap of relations, allocations and collaborations between all these initiatives
- there will be national infrastructures / contact points for electronic health records and european exchange of patient summaries for clinical care
- PGRS and GO-NL etc. are under discussion of they fit the needs for the allegedly 5% threshold that is needed to be a reference resource
- BREXIT prevented the easy addition of 500k genome data
- the undefined informed consent from pre GDPR times prevent the use of the already collected genomic data in the Netherlands
- WGS or WGE might be the way to go for the Netherlands
- Jeroen suggest to add another spread-sheet map to collect all relevant projects that all FAIR genome members are involved in to enable synergies
- Jeroen asks to collect and gather all relevant guidelines, doesn't matter how small they are
UP UNTIL NOW: Ppt slides on Github :. 2 posters @ health-ri 2020 :. GitHub page is up to date :. Agreed upon “Metadata version 1” is currently available :. Meta-data term and attribution alignment and mapping worked well :. Received feedback from health-ri poster visitors: a) the issue to define the “biological sex types” continues being a struggle, poster visitors suggested “raised as…” and “undetermined/unknown” b) patients can be renamed as “individual” c) ethnicity can be renamed sa “population” -> Esthers suggests to consider the actual use case in terms of information retrieval: biological sex is more important to health-care than gender
- Jeroen suggest to work together with Esther on this issue
- Patrick suggests that on a genomic level these information are crucial
- Jeroen concludes that there is a “context” field necessary to track the relationship of how this specific information/data was captured
- Esther remarks that these suggestions are still semantically and conceptually different and there need to be a careful consideration which terms to finally use in the FAIR genome project
- Jeroen stresses again that the context and use case of the data capture is super crucial and need to be tracked too
- Patrick remarks that this metadata in the future might not be relevant anymore, since the data will reveal it :. “Family information” remains tricky to define :. “First contact” field (date field) is open for question if it’s necessary to be captured:
- Jeroen suggest that in line of treatment, the first contact is important to capture
- Patrick remarks that some fields should not be static but rather dynamic; time-points and related field are important (longitudinal problem); he refers to transmarts and their “event”-feature that would rely on these longitudinal information :. Daphne remarks about informed consent with regards to “family information”
- discussion consensus: this information anamnestic; however reliability and authenticity of this information is questionable
:. “Sample location” vs “data location” still needs to be figured out :. Open debate “intended” vs “actual” “insert size” - “inner size”(?) suggested by Lennart instead of “intended size” - big discussion whether or not this field is necessary whatsoever
- Terry remarks that in the 1+MG context, people would search for that; agreement in the room
- Jeroen suggests that there are different use cases for information retrieval, again he points out the need for a contextual field here (e.g. clinic etc)
- Marielle suggests that there need to be quality checks
- this field could be the quality indicator
- insert size range might be a way to do it, by Joeri
- if no one would need this information, it should be scratched off, by Joeri
- Lennart suggests that this information can be super useful
- consensus that this field could be optional
:. Quick intro to ‘null flavours’: https://www.hl7.org/fhir/v3/NullFlavor/cs.html - Joeri strongly suggests to incorporate these - Jeroen comments that the peerlsnur project already uses them :. The outreach and interaction with a list of projects is under way (see presentation slide) :. Jorie mentions that illumina is working on similar things :. Metadata-model is released on github; data-model is under review - Patrick observes that meta-data and data model are mixed here
- in the guideline it should be make clear that different technical implementation are possible
NEXT STEPS: :. See presentation on Github :. Lennart suggests interoperable variant descriptions, and has 3 reference implementation projects to point to
UP UNTIL NOW: Ppt slides on Github
- variety of roles from WP members help to shape this in the right way with regards to the different point of views and needs; legal expert is missing with respect to all the peers
- machine readable IC is desired, enriched with A-DAM and DUO
- last meetings outcomes: annotated IC form based on radboud biobank consent forms
- machine readable IC will increase the findability of data and material, and clear access and consent limitations
- semantic model by Esther and Katy visualises the relationships between concepts and available terms (i.e. dublin core, ICO, DUO etc.)
NEXT STEPS:
- the selected elements will be transformed into ontology terms (DUO, ICO, etc.)
- pilot phase to try that with different consent forms
- no form will be developed, but a framework with recommended elements and their defined ontological terms
- Daphne confirms that the UMC’s are all busy with the “broad” consent implementation
- Radboud UMC has an general implementation, and specific internal pilots
- Terry suggests to include in the FAIR genome guidelines some feedback from the implementation and pilot approaches of different UMC’s
- Terry suggests that the “broad IC” part is rather on the flexible part of the IC model, and not the fixed one
- Patrick remarks that local research nurses could be facilitated to convert the analogue forms to the digital version
- if NKI and PMC (and others) would send their forms to WP4, then they could receive a machine actionable form that could evaluated and subsequently piloted
- Jeroen remarks that a legal expert should assess that the transformation from analogue to digital is sufficient enough (to hold their legal power; with regards to METC etc. in the future too) TO FACILITATE findability and not necessarily the access (= context for legal assessment)
ADDITIONAL:
- discussion about semantic model and application in this project: :. Feasibility of certain consent terms :. The future might be that all of these informations are captured on individual basis and their choice :. Esther explains: currently most patients consent on the same set of principles anyways, and individual choices are additional; use cases are relevant again :. Patrick stresses the reproducibility of science, however Esther points out that the WP4 has a specific focus :. The actual access to the data that was found via these standardised information is still to be decided by the respective DAC (data access committee) :. Lennart suggests to look at dynamic consent :. Esther explains that dynamic consent is possible with their current semantic model :. Lennart remarks that dynamic and withdrawn consent are different things :. Terry confirms that it is a good idea to have a fixed and flexible part of this IC model
What has been done: Use cases for technical solutions for reference verifications -diagnostic ngs workflow: harmonised NGS workflow HARMONISED IS NOT IDENTICAL
-interoperabl variant descriptions -datasharing Progress WP5: presented by Lennart Johanson 15:30
UP UNTIL NOW: Presentation on Github Use cases for technical solutions for reference verifications 1.diagnostic ngs workflow: harmonized NGS workflow HARMONISED IS NOT IDENTICAL GATK4 Less relevant in the verification process
…
NEXT STEPS: Have to do check individual steps of modular pipelines David asks what do you test.. Format of output? Lennart yes format of output Jeroen for guideline: test input and output You should not create a pipeline that is a black box
Hinri asks if in a not a fixed module, you always have to perform the whole pipeline, so always remove duplicates? Lennart if you want to be aligned/harmonised yes David can be done in any other way as well, lennart yes Lennart, we have to determine when it is sufficiently equal
Jeroen: NIPT be a good test case
Because of shortage of time Marielle asks 15.45 what needs to be prioritized from the agenda. Lennart states he needs only a few minutes more.
2.-harmonised variant calling, mutulizer three project HGVS to machine learning format, Because Lennart is not part of this project he does not know exactly what the content is Discussion Patrick: is this the VIC? Lennart no is not the VIC, Marielle HGVS is description of nomenclature
3.-datasharing…VARDA2
WP2&3&4 integration WP leads will make a separate appointment Connecting to daily practice (VKGL, VKGN, KMBP……) Will align with 1+MG action Next steps -Wp4 test model in nki and maxima i nformed consent -WP2/3 and 4 write a first format for guidelines (bullits) -Discuss in a zoom meeting in one month. -Jeroen will start the bullit lists -test in maximia contact Hinri test radboud jeroen van reeuwijk
Do we have a reporting obligation to ZonMW action Marielle