Traditionally, publishing the results of a genomics study has meant sharing statistical results from genome wide association studies—statistics such as effect sizes, minor allele frequencies, and p values—that are of great value to the genetics community. Recently, however, there has been interest in novel methods of sharing this data, methods that allow researchers to ask their own questions about these genomic datasets. One such effort is the Knowledge Portal network at the Broad Institute. These public portals allow researchers to, among other things, run their own custom analysis on large disease specific genomic datasets. Here we discuss work, done in collaboration with the Knowledge Portal team, trying to understand and quantify the privacy concerns raised by this new and exciting method for sharing genomic data, and discuss possible methods to address these concerns.
Back to Algorithmic Challenges in Protecting Privacy for Biomedical Data