Poster ID
P-10
Poster Title
From Silo to Standard: Responsible Discovery and Access for ZERO Childhood Cancer Data
Authors
1Kamile Taouk, 1Youjin Shrestha, 1Hugo Ludemann, 1Gawnesh Paatni, 1Angeline Bonnet Veeneman, 1,2Mark J. Cowley, 1,2Marie Wong-Erasmus
1Children’s Cancer Institute, Randwick, NSW 2031 Australia
2School of Clinical Medicine, UNSW Medicine & Health, UNSW Sydney, Sydney, NSW, Australia
1Children’s Cancer Institute, Randwick, NSW 2031 Australia
2School of Clinical Medicine, UNSW Medicine & Health, UNSW Sydney, Sydney, NSW, Australia
Abstract
Childhood cancer is the leading cause of disease-related death in children, and its rarity means progress depends on finding and combining relevant patients across cohorts and datasets. ZERO Childhood Cancer, Australia's flagship national precision medicine program, has built a rich genomic and clinical resource across paediatric cancer, whose data we aim to turn into a discovery engine for researchers. Such data too often sit in incompatible silos behind ad hoc access, limiting the statistical power to reveal mechanisms unique to paediatric cancers.
We present Insights, a web portal developed as part of the Australian BioCommons GUARDIANS program (enabled by NCRIS via Bioplatforms Australia) that makes this data both discoverable and responsibly accessible through one consistent interface, centralising the discovery, visualisation, access and governance workflow.
Discovery: researchers cross-filter cohorts by joined clinical and molecular features - patient, specimen and research study, with curated variants spanning SNVs, CNVs, structural variants/fusions and expression outliers, each with a clinical pathogenicity classification - and build virtual cohorts with a statistical overview.
Standards and implementation: Insights is built on HL7 FHIR R4 and GA4GH standards. Reportable variants are stored as FHIR Observations, materialised into flat views via HL7 SQL-on-FHIR ViewDefinitions and served through a GraphQL gateway, with variant notation aligned to GA4GH VRS. By decoupling internal curation from a standards-based, Apache-2.0 delivery layer, Insights lets organisations keep their own pipelines while presenting one interoperable interface rather than another silo.
Governance, security and privacy: a tiered model governs every record - published records are openly discoverable, while unpublished records are gated by approval status, internally visible versus requiring an explicit data-access-committee grant, enforced at query time by a fail-closed access-control service. Access requests run through REMS with signed data-use agreements, time-limited approvals with automatic revocation, and audit logs. Federated authentication uses CILogon and the Australian Access Federation; data is de-identified before load and encrypted in transit and at rest.
We share our design choices, governance trade-offs, and path to national production launch - reinforcing GA4GH's vision of a federated genomics ecosystem that accelerates discovery for children with cancer.
We present Insights, a web portal developed as part of the Australian BioCommons GUARDIANS program (enabled by NCRIS via Bioplatforms Australia) that makes this data both discoverable and responsibly accessible through one consistent interface, centralising the discovery, visualisation, access and governance workflow.
Discovery: researchers cross-filter cohorts by joined clinical and molecular features - patient, specimen and research study, with curated variants spanning SNVs, CNVs, structural variants/fusions and expression outliers, each with a clinical pathogenicity classification - and build virtual cohorts with a statistical overview.
Standards and implementation: Insights is built on HL7 FHIR R4 and GA4GH standards. Reportable variants are stored as FHIR Observations, materialised into flat views via HL7 SQL-on-FHIR ViewDefinitions and served through a GraphQL gateway, with variant notation aligned to GA4GH VRS. By decoupling internal curation from a standards-based, Apache-2.0 delivery layer, Insights lets organisations keep their own pipelines while presenting one interoperable interface rather than another silo.
Governance, security and privacy: a tiered model governs every record - published records are openly discoverable, while unpublished records are gated by approval status, internally visible versus requiring an explicit data-access-committee grant, enforced at query time by a fail-closed access-control service. Access requests run through REMS with signed data-use agreements, time-limited approvals with automatic revocation, and audit logs. Federated authentication uses CILogon and the Australian Access Federation; data is de-identified before load and encrypted in transit and at rest.
We share our design choices, governance trade-offs, and path to national production launch - reinforcing GA4GH's vision of a federated genomics ecosystem that accelerates discovery for children with cancer.
Digital Poster