First living scientific figure: articles can now keep pace with scientific discovery
22 April, 2015 | Thomas Ingraham |
|
|
Last year, open science developer-advocate Jure Triglav wrote a post on how scientific figures should work in 2014. Jure went beyond encouraging the use of scientific figures that update in real-time, he created one on his blog; a neat graph that auto-plots the number of sun-mentioning tweets in the Bay Area against actual cloud coverage data, to see if the crowd can predict good weather.
Now in 2015 we can present the first example of an auto-updating ‘living’ figure in a published, peer reviewed original research article.
After reading Jure’s post, behavioural neuroscientist Björn Brembs at the Universität Regensberg approached us at F1000Research to help create a dynamic figure that could visualize the variations in locomotor behaviour of Drosophila Canton S strains sourced from different labs. Canton S is the most frequently used wild type control strain in Drosophila experiments, and there is an unspoken assumption that these flies are effectively identical between labs.
Björn, together with his collaborator Julien Colomb at the Freie Universität and the F1000Research development team (special nod to our whizz kid developer Ravi Kumarasighe), set about coding a living figure in R that could effectively illustrate this data whilst adhering to good scientific publishing practices. The first version of their article included a static Figure 4 and a call for other researchers to contribute their findings by uploading their data to the truly living figure in Version 2 of their article (see our previous blog post and interview with the authors). Version 2 has now been published, and today the first data from another group, Gregg Roman at the University of Houston, has been added to the now fully functional Figure 4.
When a researcher submits their raw data to this figure, a principal component analysis is automatically run on all the strains and the results are plotted onto the three graphs; the closer the plots, the more similar the strains. The figure has full version control so users can see how the figure has progressed over time, every contributor is credited in the legend and each dataset is independently downloadable, citable and has its own DOI.
As of writing, seven sub-strains have been added to the figure with the eighth expected within the next few weeks (keep a lookout for future data being added). The most recent data from Dr Roman’s team do not appear to group with any of the previously added strains, providing further evidence that these hypothetically identical control flies can actually be quite different. The living figure allows us to continue building a global picture of the variation that exists in this “standard” control strain; if you maintain your own Canton S stock we implore you to test your flies (you will need to obtain the necessary hardware from Björn) and contribute your data.
This living figure will not be a one-off. Now that we have the R code environment set up on our servers, any authors who additionally submit R code alongside their article to make their figure(s) ‘living’ can easily do so (we will add coding guidelines to the site shortly). They can also submit code to enable a range of plotting or analysis options for their figure as Brembs and Colomb already did in Figure 3 of this same article (they just submitted data and the code, and some adjustable parameters). Similarly we are interested in publishing Python Notebook content within our articles which would enable readers to adjust the plotting of the data in a way they see fit.
Since then, we have met a lot of very excited researchers! The potential this opens up for all sorts of living articles containing living datasets and living analyses continues to grow. It even won one of the Vision Awards at the FORCE2015 meeting back in January. We now have a long string of projects that take the benefits of these capabilities and apply them to a range of scenarios to enable the communication of science to become faster and more informative. If you have a project that you think might benefit from this and would like to work with us to break new ground, please do get in touch!
|