1 of 36

ga4gh.org

2 of 36

​

  • Treat everyone with respect, empathy, consideration, and professionalism
  • Consider other points of view, eliminate your own biases
  • Acknowledge others’ contributions
  • Respect GA4GH and venue policies and rules.

​

Code of Ethics and Community Conduct

Help us create an inclusive environment

  • Abuse or harassment in any form, or threats thereof
  • Violating boundaries when previously communicated
  • Photographing or recording others without their consent
  • Disrespectful communication and any other behaviours that lead to hostile environment

​

The following will NOT be tolerated:

Read the full CECC on our website

​

ga4gh.org/code-of-conduct

Read the Guidelines for Respectful Behaviour

​

bit.ly/ga4gh-respectful-

engagement

Have questions or want to report? Contact Safe GA4GH Officer Paula Brantner at

​

conduct@ga4gh.org

ga4gh.org

3 of 36

Introductions

Ian Fore, Mallory Freeberg, and all

ga4gh.org

4 of 36

Agenda note

Soft stop at 12:00 PDT (GA4GH Steering Committee)

Continuing until 12:30 PDT

ga4gh.org

5 of 36

Section 1.1 - Review Objectives

Mallory Freeberg, Ian Fore

ga4gh.org

6 of 36

Session Objectives

  • Identify ongoing interest in issues related to cohort building
  • Identify potential scientific use cases for follow up
  • Future collaborative activities?
    • Potential GIF Project?
    • Hackathons/codeathons

ga4gh.org

7 of 36

Section 1.2 - Summary of previous computable cohort discussions

Ian Fore

ga4gh.org

8 of 36

Computable Cohort Hackathon 2022

  • Main GA4GH groups involved
    • Clinical and Phenotypic Data Capture
    • Discovery
      • Beacon 2.0
      • Data Connect
    • Federated Analysis Systems Project
      • Cross workstream integration
      • Hackathon - trying things out
  • Asthma as a use case
    • Aggregating patients and data from multiple sources

ga4gh.org

9 of 36

Computable Cohort Hackathon 2022 - topics covered

  • Data Access - “DARathon”
    • Practical assistance on how to request data access
  • Exploration of Beacon 2.0
    • Cineca example
  • Data Connect
    • Capability of Data Connect to represent Cineca synthetic data
    • Asthma mapping example
  • Clinical Query Language (FHIR)
    • Use of terminology mappings

​

ga4gh.org

10 of 36

Workflow

https://github.com/BRCAChallenge/federated-analysis

https://pubmed.ncbi.nlm.nih.gov/35373174/

ga4gh.org

11 of 36

Federation of the VUS workflow via GA4GH WES

Jupyter

​

​

​

​

​

​

​

Genomic Data Commons - PanCanAtlas

TCGA Germline variants

Huang et al

Cancer Genomics Cloud WES

TOPMed COPDGene

BioDataCatalyst WES

DRS

DRS

WES

WES

DRS

DRS

Jupyter

​

​

​

​

​

​

​

Jupyter

​

​

​

​

​

​

​

Kids First

Osteosarcoma

Cavatica WES

DRS

WES

DRS

12 of 36

Merged analyses

13 of 36

GA4GH Implementation Forum (GIF)

  • Successor to Federated Analysis Systems Project
  • Supports GA4GH shift to focus on Implementation
  • GIF Projects
    • Three or more data sources
  • Is there a potential GIF project
    • Aggregate data from EGA, JGA, dbGaP, HDR(UK), Other
    • Local compute as needed

ga4gh.org

14 of 36

Section 1 - Identify existing sources from which cohorts could be assembled

Speaker name

ga4gh.org

15 of 36

HDR UK

HDR UK is the UK’s national institute for health data science.

​

The HDR UK mission is to unite the UK’s health and care data to enable discoveries that improve people’s lives. This is done by uniting, improving and using health and care data as one national institute.

​

https://www.hdruk.ac.uk/

​

Speaker: Tim Beck (virtual)

ga4gh.org

16 of 36

A journey towards cohort building and beyond

Biobank participant discovery

  • BBMRI UK
  • UKCRC TDCC

​

COVID-19 cohort building

  • CO-CONNECT
  • BC Platforms technology

HDR UK cohort discovery tool

  • Available from the HDR Innovation Gateway
  • ~5M individuals discoverable across 22 datasets

HDR UK federated analytics

  • Open standards/technologies
  • Enabled within a national TRE ecosystem

Speaker: Tim Beck (virtual)

ga4gh.org

17 of 36

Cohort discovery: datasets

Speaker: Tim Beck (virtual)

ga4gh.org

18 of 36

Cohort discovery: searchable variables

Speaker: Tim Beck (virtual)

ga4gh.org

19 of 36

Cohort discovery: query builder

Speaker: Tim Beck (virtual)

ga4gh.org

20 of 36

TRE-FX: Federated network of TREs to enable safe analytics

Speaker: Tim Beck (virtual)

ga4gh.org

21 of 36

TRE-FX: Federated network of TREs to enable safe analytics

Speaker: Tim Beck (virtual)

ga4gh.org

22 of 36

GA4GH standards under consideration

  • Beacon v2 to be used alongside BC Platforms technology
    • Open-source version of proprietary technology

​

  • Task Execution Service (TES) use for batch execution tasks

​

  • Passports to support data access request processes

​

Speaker: Tim Beck (virtual)

ga4gh.org

23 of 36

European Genome-phenome Archive (EGA)

Speaker: Mallory Freeberg (in-person)

ga4gh.org

24 of 36

European Genome-phenome Archive (EGA)

  • What dataset content comes from your resource
    • Genomic data: un/aligned sequences (FASTQ, CRAM, BAM), variants (VCF), microarrays (various)
    • Phenotypic data: high-level public phenotypes (e.g. sex, disease), individual-level private phenotypic and/or clinical data (e.g. diagnosis information, ancestry)
    • No patient health record-like data
  • What is the interest/approach to cohort building and federated analysis
    • Discovery: EGA Beacon v2 to support finding relevant datasets (new)
    • Access: per-dataset access requests to DACs (current)
    • Download: one-time and on-demand data download (current)
    • Analysis: No plans for EGA analysis environment
    • Example: CINECA & Federated data analysis learning pathway
    • EGA & GA4GH

Speaker: Mallory Freeberg (in-person)

ga4gh.org

25 of 36

European Genome-phenome Archive (EGA)

  • Issues that would need to be addressed, e.g. operating as a Trusted Research Environment
    • Technical
      • Individual steps are technically possible; ways to streamline?
      • No analysis environment provided
    • Operational
      • Data/metadata harmonisation likely required
      • Different expectations for data access requirements and timelines
    • Legal
      • Per-dataset restrictions on where data can be downloaded to
      • Per-dataset requirements for reporting results
      • Increased risk of re-identifiability?
  • Any GA4GH (or other) standards identified as relevant?
    • Beacon, file formats, phenopackets, Crypt4GH, htsget, DUO,

Speaker: Mallory Freeberg (in-person)

ga4gh.org

26 of 36

Database of Genotypes and Phenotypes (dbGaP)

  • What is the interest/approach to cohort building and federated analysis
  • What dataset content comes from your resource
    • 2,500 studies - diverse
    • Subject and sample attributes
      • 85% of studies have simple 4-5 table model
      • Long tail of study specific models
        • Variable numbers of additional columns
    • Simple data dictionary defines table columns
      • Submitter originated

​

Speaker: Ian Fore (in-person)

ga4gh.org

27 of 36

Issues that would need to be addressed - dbGaP

  • Control of data access is important - but managed
    • Common infrastructure for requests
    • Data access committees make decisions
  • Diversity of data
    • The diversity is genuine - reflects:
      • Domain specific attributes
      • investigator originated study design
      • ​

Speaker: Ian Fore (in-person)

ga4gh.org

28 of 36

dbGaP - relevant standards

GA4GH Data Connect

GA4GH DRS

FHIR

ga4gh.org

29 of 36

Data aggregated from dbGaP studies using

GA4GH Data Connect

​

Key elements:

  • User provided data dictionary
  • Verified mappings

ga4gh.org

30 of 36

JGA / DDBJ JGA: Japanese Genotype-phenotype Archive

  • What dataset content comes from your resource
    • WGS/WXS and other array/sequencing data
      • 355 research entries (477 datasets)
      • Many from cancer genomics (list)
    • Maintained in the DDBJ’s supercomputer system
      • Providing computing resources to the users
  • Any GA4GH (or other) standards that you have identified as relevant
    • WES/TRS for enabling 'code to data' analysis
      • Collaborating with the ELIXIR cloud team through BioHackathon for interoperability of our WES implementation
    • VCF and other file formats
    • Passports for user authentication

Speaker: Tazro Ohta

ga4gh.org

31 of 36

JGA / DDBJ JGA: Japanese Genotype-phenotype Archive

Speaker: Tazro Ohta

ga4gh.org

32 of 36

JGA / DDBJ JGA: Japanese Genotype-phenotype Archive

  • What is the interest/approach to cohort building and federated analysis
    • Bulk data analysis: create a secondary database
    • Benchmark platform: allow users to perform benchmarking over private data
    • Curated list of private data across the dbGaP/EGA/JGA
  • Issues that would need to be addressed, e.g. operating as a Trusted Research Environment
    • How can we provide sufficient metadata for the user search to identify the data of their interests?
    • How can we trust the incoming code?
      • We have a review system on our TRS but depending on manual effort
    • How can we make it easy to do bulk data use application for each site?

​

Speaker: Tazro Ohta

ga4gh.org

33 of 36

Section 2 - Identify candidate scientific questions that could be answered across implementations

Speaker name: TBD

ga4gh.org

34 of 36

Scientific question #1: Asthma

Speaker: Ian Fore

ga4gh.org

35 of 36

Scientific question #2: Suggestions?

  • Alzheimer’s / Parkinson’s
  • Healthy elderly cohort

Speaker: Group

ga4gh.org

36 of 36

Section 3 - Wrap-up, next steps, & actions

Speaker name: Group

ga4gh.org