Barriers To Data Sharing

I’ve made this point about the role of funders before, but Fiona Nelson is absolutely right (boldface mine):

Genomics is a data science. The human genome is vast – 3 billion base pairs. To make any significant findings from the data, you need lots of data for validation and comparison. It is impossible for any one research group, lab or institute to generate all the data that is relevant for any one disease, so data sharing and collaboration across institutions is paramount for the quality of genomics research….

While virtually all of the research community recognise the need for data sharing and collaboration, it is an uphill battle to change the culture and incentive structure of academia. The most recognised metric for research output number of papers published in high profile journals. As long as data sharing is not part of the agenda or incentive structure, it remains a side project – a nuisance – which is only taken care of last minute if at all when publishing papers. Researchers would rather spend their time writing papers and grants than spending time and effort making their data available and accessible….

I think all researchers who are using biomedical data should think twice about the impact of the data. The individuals, often patients with serious diseases, gave consent for their samples and data to be used for research to make a difference for research to help future patients. Not sharing data, not making the most of data, is not meeting the expectations of the data donors. So even if it takes time and effort, it is in my opinion, part of the obligation of the research community to enable data sharing to maximise research impact. Next to incorporating this mindset among researchers, I think the power to change the incentive system lies with the research funders. The day that funders require that all your biomedical research data must be made available for reuse by the research community before they will give you another research grant, I am sure that you will start taking data sharing seriously.

Funders not only need to mandate data release plans–and make them scorable parts of all grants, they also need to compensate scientists for this. Which is to say, this must be an item (even if only a small one) in budgets.

This entry was posted in Data Stuff, Funding, Genomics. Bookmark the permalink.