The Global Alliance for Genomics and Health was thrust into the headlines last week when Google ($GOOG) joined its ranks. This week it met for the first time to discuss its goals. Development of an open-source public application programming interface (API) for sequencing reads and genetic variants is high on the agenda.
Speaking to Nature, Broad Institute deputy director and chair of the Alliance's steering committee David Altshuler outlined the thinking behind the API. Interoperability has come on since the early days of sequencing--when British and U.S. teams collaborating on the Human Genome Project struggled to interchange data--but the siloing of information remains a problem. The Alliance plans to overcome these challenges by shifting the focus away from interoperable file formats and on to an API.
"There are certain kinds of human activities, such as the World Wide Web, where information being in an interoperable framework with appropriate security and privacy has really transformed things in a very positive way. We're not in that world yet for genomics," Altshuler said. The API is one step towards establishing this framework for genomics. Google is playing a role--it unveiled an API when it joined the Alliance--but Altshuler stressed the tech giant is just one of 151 partners on the project.
The 151 members span biopharma, academia, technology and disease advocacy, with Illumina ($ILMN) and Merck ($MRK) sitting alongside Google as some of the highest-profile partners. The breadth of the alliance is testament to the far-reaching implications of genomics and the range of skills needed to fulfil its potential. Sheer scale is important too. "In order to learn from the data, you need very large data sets, larger than any entity can collect on its own," Altshuler said.
While the Alliance was discussing the benefits of massive data sets in one part of London, the fallout from the botched implementation of the English health records system was still rocking the political part of the city. Altshuler warned of the need to strike the right risk-benefit balance on such projects. "If one takes the view of only focusing on the harms, then there are approaches to put data behind high walls so no harms can occur and no learning can occur," he said.
- read the Nature Q&A