Skip to main content
Fig. 1 | Genome Biology

Fig. 1

From: sPLINK: a hybrid federated tool as a robust alternative to meta-analysis in genome-wide association studies

Fig. 1

Comparison of sPLINK (c), aggregated analysis (a), and meta-analysis (b) approaches: Aggregated analysis requires cohorts to pool their private data for a joint analysis. The meta-analysis approaches aggregate the summary statistics from the cohorts to estimate the combined p-values. In sPLINK, the cohorts calculate the model parameters (M) from the local data and global model, generate noise (N), and make the parameters noisy (M′) in an iterative manner. The aggregated noise and noisy parameters are in turn aggregated to update the global model or build the final model. sPLINK combines the advantages of the aggregated analysis and meta-analysis, i.e. robustness against heterogeneous data and enhancing the privacy of cohorts’ data. Yellow/blue color indicates case/control samples

Back to article page

Genome Biology

ISSN: 1474-760X

Contact us