Publishing ENCODE: The Use of Interactive Graphics for Navigation and Exploration of Research

Publiceret Juli 2014


The ENCODE (Encyclopedia of DNA Elements) project, funded by the National Human Genome Research Institute, endeavors to describe the functional elements of the human genome. In September 2012 Nature published a package of research from the ENCODE consortium, including 6 papers from Nature and 24 from two other journals, all released simultaneously in a new online format that included an interactive graphic interface for exploration and navigation (Figure 1).

Figure 1. The encode website with explorer on first launch.

Threads: A New Content Type

In addition to publishing the traditional research papers from ENCODE, Nature introduced a new concept in publishing called ‘threads'. Threads are articles that compliment a set of research papers by highlighting topics that are only otherwise covered in subsections of individual papers. Each thread consists of relevant paragraphs, figures and tables from across the set of papers, united around a specific theme, such as DNA methylation or machine learning.

Threads were created to help researchers cope with the enormous amount of information produced during the five years of the ENCODE project to date. The aim was to help scientists more easily focus on their area of interest. The content of thread articles was selected by the ENCODE consortium. Threads are meant to flow as a narrative that tells an alternative story to the papers themselves, and are not merely a collection of disjointed ‘research highlights' (Figure 2).

Figure 2. A thread page.

Key Publishing Questions and Goals

Threads presented Nature, as a publisher, with some interesting questions, such as: Where will threads ‘live'? (unlike papers, they do not have a natural habitat within a journal); How will users locate them?; And how can we present threads so that their relationship to papers are quickly and easily understood? As art director for Nature, I set about to find a solution to these and other questions, along with Nature editor Magdalena Skipper and information designer Max Gadney.

Before publishing ENCODE, Nature had previously presented collections of research in a basic linear format online, with research articles and other related content traditionally presented in list form (Figure 3). With the ENCODE project, we decided to try something new.

Figure 3. A web focus of mouse genome.

Our goal was to create a home for the collection that would: Show a holistic display of the collection at a glance; Introduce threads and show their relationships to papers; And serve as a navigational tool for the collection, with direct links to the papers and thread documents.

The Nature ENCODE Explorer

Our solution was a web portal with an interactive tool for visualization called the Nature ENCODE Explorer, with the aim of properly introducing the concept of threads by showing their relationship to papers by displaying connections in a clear visual format. The tool also displays the entire collection in one view, and allows researchers to navigate to both papers and threads from the interface itself (Figure 4). For those who might prefer a more traditional experience, we provided a simple list of papers and threads further down the web page.

Figure 4. This shows explorer opened and showing a connection.

Our new design used classic visualization principles to create a ‘map' of the global ENCODE research landscape, with threads on the left side of an ellipse and papers on the right. Connections between papers and threads were made explicit by the presence of a connector with fluid lines (a nod to a physical thread, or string). Papers were given the form of icons that look like physical papers, and threads were represented as colored numbered circles, to differentiate them from the research papers.

Creating this visualization in a digital format allowed for interactivity. One click on a thread shows two things at once: its relationship to individual papers in the collection, and a summary of the thread with a live link to the thread document. A click on a paper icon shows the title, author, journal, and DOI, with a live link to that paper, and related thread icons that are also clickable. The design ulitizes an open network, where one can flow seamlessly between the thread and paper sides, and a roll-over state for thread and paper icons for quick scanning of the collection ­(Figure 5).

Figure 5. A fully expanded Explorer page.


Anecdotal evidence suggests that both the threads and the Explorer itself were well received in the research community. Unfortunately, due to the speed and complexity of publishing such a large collection in cooperation with other publishers, we were unable to build a way to measure actual metrics for the interactive portion of the site (such as page views or time on page).

Nevertheless, judging by the interest that the Explorer generated, it is safe to say that the community welcomes efforts by publishers to create tools that help sift through large collections of research. Design and visualization will continue to play a key role in these efforts going forward.