The ability to detect an individual's participation within a large scale genetic association study was proven in the past year, raising privacy concerns and leading to higher restrictions on such experimental datasets. A new study this week in Nature Genetics reports an improved method to infer membership from these datasets, and suggests increased restrictions may be needed.

Kevin Jacobs and colleagues used an extension of a method originally reported in 2008 to detect an individual or close relative's membership within a genome-wide association study. The new method shows increased power to detect an individual's participation within the study, and extends the range of summary statistics that may allow such detection. They are also able to predict whether the individual was a case ― with the relevant disease ― or control in the study. While the authors estimate the lower bound of the power to detect membership from these datasets using this method, other methods may be able to show higher power. This study highlights the importance of revisiting privacy concerns over how much information from these studies should be made publicly available.