MOLGENIS Armadillo: a lightweight server for federated analysis using DataSHIELD

Area of Science:

• Bioinformatics
• Computational Biology
• Health Informatics

Background:

• Large-scale human health data from cohort studies, registries, and biobanks are crucial for identifying lifecourse risk factors.
• Combining these diverse data sources enhances statistical power, aids in detecting rare outcomes, and allows for result replication.
• Traditional data integration methods involve data transfer or pooled analyses, which present ethical, legal, and time challenges.

Purpose of the Study:

• To introduce MOLGENIS Armadillo, a lightweight server designed to simplify the implementation of federated analysis solutions.
• To provide data owners with an accessible tool for installing federated infrastructure and managing users and data.

Main Methods:

• Federated analysis enables remote data analysis without sharing individual-level data, addressing privacy and logistical concerns.
• MOLGENIS Armadillo supports federated analysis solutions like DataSHIELD.
• The system is implemented using open-source R packages ('MolgenisArmadillo', 'DSMolgenisArmdillo') and a Java application ('ArmadilloService'), available via CRAN and GitHub.

Main Results:

• MOLGENIS Armadillo provides a user-friendly server for establishing federated analysis environments.
• The implementation leverages existing open-source tools and packages for broad accessibility.
• Facilitates secure, remote analysis of sensitive human health data.

Conclusions:

• MOLGENIS Armadillo streamlines the adoption of federated analysis, overcoming barriers associated with traditional data sharing.
• It empowers researchers to leverage extensive health datasets securely and efficiently.
• This promotes advanced epidemiological research and discovery by enabling collaborative analysis of distributed data.