myExperiment: sharing scientific workflows through social networking
Publication date: September 2013
Summary
The Jisc-funded Virtual Research Environment (VRE) project, myExperiment, designed a social networking site where researchers can share their research processes (workflows) and find collaborators. Six years after it launched, the site hosts more than 3,000 scientific workflows and has at least 9,000 registered users worldwide who are organised into around 300 subject groups. Ongoing maintenance and development are funded by major EU and UK research council projects which are using it as a virtual research environment. Researchers are posting a wider variety of experimental information than originally envisaged, prompting interest in the platform as a novel means of scholarly communication.
What myExperiment has facilitated
The sharing of scientific workflows for reuse and re-purposing by researchers across a wide range of disciplines and for training. The use of workflows has led to new scientific discoveries, especially in the field of bioinformatics.
Interest in a new paradigm for scholarly communication.
Important lessons for VRE developers
Build your user base and your VRE at the same time. Make sure you have lots of users and be guided by their requirements.
Adopt or adapt a 'look and feel' that users are already familiar with e.g. Facebook.
Pay attention to security and users' concerns about sharing intellectual property. This is the main reason why popular services such as Facebook are not suitable for research use.
The challenge
The idea for myExperiment grew out of an international project to combat sleeping sickness (trypanosomiasis) in African cattle. Researchers wanted to track down the genetic root of resistance to the disease by analysing data collected by multidisciplinary teams held in databases all over the world. Taverna, which had been developed at the University of Manchester, enabled them to create workflows to automate the processes needed to query the databases sequentially, rather than having to enter results from one query into the next database by hand.
The researchers began to share their workflows, often by e-mail or by posting them on websites or wikis. This gave Professor Carole Goble, whose team had developed Taverna, and Professor Dave De Roure, then at the University of Southampton, the idea to build a social environment, similar to Facebook, where workflows (rather than photos and music) could be shared. Researchers would upload their workflows for others to download and reuse directly, or edit to create a new workflow, perhaps striking up a collaboration with the original author or others. Jisc funded the MyExperiment VRE project to develop this social environment.
The innovative approach
myExperiment was designed by adopting Web 2.0 principles for the benefit of digital research. A major challenge was to build an interface that would give researchers control over how their workflows are shared. “I'd underestimated one task: getting the interface right with respect to credit, attribution and licensing. These matter to scientists and is why myExperiment is different from Facebook,” says De Roure. Within the first year, the number of users grew rapidly, indicating that researchers were happy with the way these issues are dealt with on the site. As a consequence, scale-up work had to proceed much sooner than expected.
myExperiment is also unusual in its focus on sharing research processes or methods. “Everyone is talking about data, but we're talking about what you do with the data,” says De Roure.
The results
An early success was the unexpected finding that high blood cholesterol confers resistance to sleeping sickness in cattle by enabling them to survive the high levels of inflammation caused by the disease. By automating search processes with Taverna workflows - honed by collaboration facilitated by myExperiment - bioinformaticians were able to search data about all genes remotely associated with resistance to the disease. Without these tools, the task would have been too onerous and time-consuming to contemplate.
Six years after it launched, myExperiment has more than 9,000 registered users organised into more than 300 'communities' of interest. Disciplines covered range from astrophysics to biodiversity to text mining.
In response to user requests, myExperiment added the facility to share 'packs' of information related to workflows, such as results of experiments, research data or published papers. Packs are entities in themselves that can be tagged, reviewed and shared. The development of myExperiment has also led to other new sites, for example BioCatalogue, an index of web services in the life sciences making them far easier to discover and link to workflows.
Follow-on funding from Jisc enabled myExperiment to take an open linked data approach to the repository of workflows and packs it was now building up. This meant that the repository could be coupled seamlessly with other repositories taking a similar approach. In parallel, the myExperiment team started to evolve the idea of 'packs' into the more sophisticated concept of 'research objects'. These are collections of information that support re-playable, repeatable, reproducible, reusable, re-purposable and reliable research.
“We've innovated in what can be shared and this has attracted the interest of people working in research communication and data sharing,” says De Roure. Research objects could complement, or even replace, traditional scholarly communication based on the published, peer-reviewed journal article. Work supported by the EU's 7thFramework programme is exploring this possibility by investigating a successor to myExperiment that will enable the sharing of a wide range of research data and other material across a range of disciplines.
Sustainability
Since start-up funding ended, myExperiment costs have come from research projects that have needed its facilities. These include Biodiversity Virtual e-Laboratory, which is building a virtual research environment for biodiversity researchers who use workflows to analyse distributed data; and SCAPE, which is developing workflows for the preservation of very large amounts of data in repositories. Both are EU 7th Framework projects.
myExperiment was originally hosted at the University of Southampton, but moved to a commercial provider to guarantee more network support than the partner universities could provide at the time. It is now hosted by the University of Oxford where De Roure has become director of the Oxford e-Research Centre. A future option is to move it to the 'cloud'. “We didn't investigate this option earlier because we wanted to keep all the innovation and content on one site, but we're looking at it now. We may go virtual, then we could run multiple myExperiments,” he says.
Further information
