Genomic crowdsourcing with privacy

See allHide authors and affiliations

Science  08 Jun 2018:
Vol. 360, Issue 6393, pp. 1083-1084
DOI: 10.1126/science.360.6393.1083-f

Privacy concerns can be an obstacle to large-scale donation of the genetic material that is essential for understanding how genetic variants are associated with disease. Cho et al. have developed an approach that has the potential to be scalable to as many as a million genomes. Their system is based on dividing genotype and phenotype data from each individual among multiple servers in such a way that no one server can infer the original data. An attacker would have to hack all of the servers to extract the data. Computational approaches that simplified corrections for misleading correlations and sped up the system were used to reproduce three published genome-wide association studies representing 23,000 genomes and could make the process cost-effective.

Nat. Biotechnol. 10.1038/nbt.4108 (2018).

Stay Connected to Science

Navigate This Article