Secure and federated quantitative trait loci mapping with privateQTL.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Understanding genotype-phenotype relationships is key for personalized medicine.
• Expression quantitative trait loci (eQTL) mapping links genetic variants to gene expression.
• Current eQTL mapping faces limitations in statistical power and data privacy.

Purpose of the Study:

• To introduce privateQTL, a novel framework for secure and federated eQTL mapping.
• To address privacy concerns in multi-site eQTL studies.
• To improve accuracy and scalability in genetic association studies.

Main Methods:

• Leveraging secure multi-party computation (SMC) for privacy-preserving data analysis.
• Implementing a federated learning approach for distributed eQTL mapping.
• Comparing privateQTL performance against traditional meta-analysis methods.

Main Results:

• privateQTL demonstrated superior performance compared to meta-analysis in a real-world scenario.
• The framework accurately corrected for covariates and batch effects.
• Achieved higher accuracy and precision in eGene-eVariant mapping and effect size estimation.
• Showcased modularity and scalability for diverse applications.

Conclusions:

• privateQTL offers a practical and effective solution for privacy-preserving collaborative eQTL mapping.
• The framework is adaptable for various molecular phenotypes and large-scale genomic studies.
• Enables robust genetic association studies while safeguarding sensitive individual data.