Notice
- Important: This guidance is under active development by NHS England and content may be added or updated on a regular basis.
- This Implementation Guide is currently in Draft and SHOULD NOT be used for development or active implementation without express direction from the NHS England Genomics Unit.
Data Storage Components
The UGR contains three primary classes of data, each necessitating a distinct approach to data management and storage:
- Structured data that captures the details of genomic test orders, sample processing, bioinformatics analyses, test reports and reporting etc. This data is characterised as being complex, highly structured based on HL7 FHIR standard, but low in volume compared to the primary genomic data.
- Unstructured data held in other general file formats such as CSV and PDF to support legacy systems incapable of consuming structured data.
- The large-scale data generated by DNA genotyping and sequencing technologies, such as the primary sequencing reads (in SAM, BAM or CRAM formats) and derived data such as variant calls (in VCF) and other data produced by bioinformatics analyses.
To accommodate this diversity and maximise functionality, multiple data repositories are utilised rather than a single physical repository. The Diagram below shows the separation components required to support the UGR.
Central FHIR Store
It is expected that all FHIR resources will be stored in a national FHIR store, available through RESTful API Access via the national API platorm.
Federated Object Store
Not all genomics-related data is suitable for storage within FHIR repository. To accommodate different data classes, multiple Federated Object Stores can be utilised, leveraging cloud object stores. To ensure consistency, the GA4GH Data Repository Service (DRS) protocol will be adopted as the standard retrieval mechanism for all data accessible via the UGR. The use of DRS enables the UGR to support both centralised and federated storage models, or a combination thereof. DRS URIs stored within FHIR DocumentReference Resources for each patient will serve as the logical identifiers for all files hosted within the Federated Object Stores.
Bulk Query Interface
For population health management and research use cases, the UGR requires the ability to perform queries across larger sets of data. The APIs supporting these queries are typically read-only and may operate asynchronously to facilitate population-level analysis. The Bulk Query Interface component is designed to support these use cases, with the data made available potentially being transformed and reformatted to optimise bulk querying an RDMS source.