Privacy-preserving and efficient validation of results in genome-wide association studies (GWAS)

The study proposes a privacy-preserving framework that validates the correctness of genome-wide association study (GWAS) statistical test results while protecting individual privacy. The researchers publicly release the workflow, outputs, and partially noised metadata satisfying local differential privacy; verifiers combine this with another statistical study on public datasets to distinguish correct from incorrect results. Evaluation on real genomic data shows that summary statistics of only a small number of variants can also validate with high accuracy, and the additional privacy risk of the metadata does not increase beyond the risk already inherent in sharing research results.