Poster ID
P-17
Poster Title
Implementing the GA4GH Task Execution Service Standard for Federated Analytics across UK and Singapore Trusted Research Environments
Authors
Tim Beck – University of Nottingham, UK
Hazel Lockhart – Jones – Swansea University, UK
Vasiliki Panagi – University of Nottingham, UK
Justin Biddle – Swansea University, UK
Jonathan Couldridge – University of Nottingham, UK
Jonaa Eva – University of Nottingham, UK
Alex Lee – Swansea University, UK
Tri Thien Nguyen – University of Nottingham, UK
Andy Rae – University of Nottingham, UK
Low Pin Yan – MOH Office for Healthcare Transformation, Singapore
Xing Yi Woo – Agency for Science, Technology and Research (A*STAR), Singapore
Simon Thompson – Swansea University, UK
Philip Quinlan – University of Nottingham, UK
Hazel Lockhart – Jones – Swansea University, UK
Vasiliki Panagi – University of Nottingham, UK
Justin Biddle – Swansea University, UK
Jonathan Couldridge – University of Nottingham, UK
Jonaa Eva – University of Nottingham, UK
Alex Lee – Swansea University, UK
Tri Thien Nguyen – University of Nottingham, UK
Andy Rae – University of Nottingham, UK
Low Pin Yan – MOH Office for Healthcare Transformation, Singapore
Xing Yi Woo – Agency for Science, Technology and Research (A*STAR), Singapore
Simon Thompson – Swansea University, UK
Philip Quinlan – University of Nottingham, UK
Abstract
Sensitive health data cannot be freely shared across borders, yet global research increasingly demands multi-jurisdictional analysis. Trusted Research Environments (TREs) enable secure analysis of sensitive data by bringing researchers to the data rather than moving data to researchers, but their heterogeneous architectures and governance models create significant technical barriers to federated analytics across sites.
We present Five Safes TES: a standards-based architecture and open-source reference implementation for secure federated analytics across heterogeneous TREs, aligned with the Five Safes Framework. The GA4GH Task Execution Service (TES) API standardises the submission, management and monitoring of analytic tasks. Three logical layers underpin the architecture: a Submission and Orchestration Layer, external to the TRE, captures analytic intent and coordinates execution; a Controller Layer, inside the TRE, enforces local governance and disclosure control; and an Analysis Execution Layer runs approved analytics against sensitive data using TES-compliant engines (Funnel or TESK). This separation allows each TRE to retain full governance autonomy while participating in federated analyses, without requiring inbound network access or modification to local infrastructure.
We demonstrate real-world deployment in two case studies. Within Dementias Platform UK, Five Safes TES powers the federated analytics service, enabling multi-cohort dementia research across UK data providers. In a transcontinental deployment, we connect two independent TREs in Singapore, at A*STAR and enTRUST, to dedicated Submission and Orchestration Layers hosted in Nottingham, United Kingdom. Each Singaporean TRE runs its own Controller and Analysis Execution Layers against OMOP CDM-formatted data, communicating outbound only, with disclosure-controlled outputs returned to the UK. This UK-Singapore deployment demonstrates genuine multi-jurisdictional federated analytics, executed without centralising control or moving sensitive data across borders, and shows how GA4GH TES can bridge distinct national and institutional data governance regimes.
By aligning open technical standards with established governance frameworks, Five Safes TES shows how federated analytics can become a routine, deployable research capability rather than a bespoke technical exercise, supporting international collaborations between research infrastructures.
We present Five Safes TES: a standards-based architecture and open-source reference implementation for secure federated analytics across heterogeneous TREs, aligned with the Five Safes Framework. The GA4GH Task Execution Service (TES) API standardises the submission, management and monitoring of analytic tasks. Three logical layers underpin the architecture: a Submission and Orchestration Layer, external to the TRE, captures analytic intent and coordinates execution; a Controller Layer, inside the TRE, enforces local governance and disclosure control; and an Analysis Execution Layer runs approved analytics against sensitive data using TES-compliant engines (Funnel or TESK). This separation allows each TRE to retain full governance autonomy while participating in federated analyses, without requiring inbound network access or modification to local infrastructure.
We demonstrate real-world deployment in two case studies. Within Dementias Platform UK, Five Safes TES powers the federated analytics service, enabling multi-cohort dementia research across UK data providers. In a transcontinental deployment, we connect two independent TREs in Singapore, at A*STAR and enTRUST, to dedicated Submission and Orchestration Layers hosted in Nottingham, United Kingdom. Each Singaporean TRE runs its own Controller and Analysis Execution Layers against OMOP CDM-formatted data, communicating outbound only, with disclosure-controlled outputs returned to the UK. This UK-Singapore deployment demonstrates genuine multi-jurisdictional federated analytics, executed without centralising control or moving sensitive data across borders, and shows how GA4GH TES can bridge distinct national and institutional data governance regimes.
By aligning open technical standards with established governance frameworks, Five Safes TES shows how federated analytics can become a routine, deployable research capability rather than a bespoke technical exercise, supporting international collaborations between research infrastructures.
Digital Poster