Beyond Bermuda: Policies to Promote Data Sharing

Daily data release was a signature feature of the publicly funded Human Genome Project 19996-2003. An open science ethos emerged from model organisms (esp. nematode) and from software and networking, and it has contended with more proprietary and traditional data-control norms and practices throughout the history of genomics. The policy of daily data release was famously decided at a 1996 meeting in Bermuda, 21 years ago this month. Today, understanding genomic variation requires sharing data beyond individual institutions, but practices and norms vary widely. The problem is even more complicated than in 1996, because uses are not just scientific, and not just about a human reference sequence. Users of genomic data include consumers interested in ancestry and doing their own exploration. Data flow come from and is used by clinical laboratories and clinicians as much or more than researchers. The data pertain to people, whose privacy and other interests must be respected, and whose informed consent is essential. We are not in Bermuda anymore. Rather than the 50 people from 5 nations represented in Bermuda, the Global Alliance for Genomics and Health has over 800 members from 70 countries, with more sure to join. Commercial firms are vastly more numerous, heterogeneous, and fully integrated into the practice of genomics. Rather than a handful of genomic startups in 1996, dozens of firms with highly diverse products and services now have genomics as part of their business, ranging from instruments to data management to software to clinical testing and integration into medical records. Thus the sources of data are more varied and numerous; national boundaries and national laws must be respected; concerns about privacy, confidentiality, and informed consent must be addressed; users are not just scientists but also consumers and clinicians; and the story will play out in a more complicated commercial landscape The challenges of constructing a medical information commons are daunting. But the norms of open science are powerful and the design principles of data access and transparent analysis can be incorporated into a data-sharing ecosystem. Doing so from the beginning is much more likely to work than imposing open science principles once business models and academic practices have become established and begun to ossify.