The HIV Vaccine Data Connector is a web tool that empowers researchers to discover connections among data from multiple experiments, labs, and research networks. For the first time, researchers can discover details of others’ work and visually compare results to find patterns and generate new insights. It is a flexible, interactive workspace where public sharing is encouraged to accelerate science, but private data are protected to preserve credit and validity. It has the potential to change the way researchers work together to battle one of the most dangerous diseases of our time – HIV/AIDS.

In 2010, 31 million adults and 3 million children were living with HIV/AIDs, up 17% from 2001. That prompted UNAIDS to demand “faster, smarter, better” programs to contain and control the disease. Traditional approaches to a vaccine have failed: unlike many other diseases, a solution won’t come from one person with a great idea because HIV is too complex and mutates too quickly. To accelerate vaccine development, isolated labs and research networks should be able to access and compare each other’s data to find patterns of immune response. But researchers often only have access to their own results and collaboration is stymied:

• Scientists and funding groups are protective of their data to ensure they get credit and that others use their ideas in a valid way.
• Labs need help to interpret each other’s data and make valid comparisons.
• Differing standards create multiple names for the same things.

In the past, we have worked on projects about open government, personal analytics, preventative health, and other areas that are being revolutionized by shared access to Big Data. As we learned about HIV vaccine research, we saw that data “siloes” are delaying a cure. HIV, vaccine components, and immune responses are terribly complex. Brilliant minds are working on their own ideas but cannot explore novel interpretations and hypotheses about the relationship of their data to those produced by other labs. The traditional model is to develop an idea and then get funding that supports access to data and pays for a central expert to do analysis. This sets a high bar and only a small portion of relationships between results are tested.

We set out to pay the costs of combining and aligning data upfront so that these incremental costs do not have to be paid to test every idea. We also set out to make all data available for comparison and discovery to every researcher contributing data. There is no limit to the valuable ideas that could be explored with open access by the whole community of researchers. Many may not pan out, but others could lead to important discoveries that could not otherwise be made. This point of view represented a major shift, even from the expectations of our project partners. We are pleased to have built consensus, even if brand new data need to be kept private at first.

One critical design choice was how to organize data. A key insight led us to pivot data on the vaccine study participants. Selecting people of interest based on one factor preserves all other experimental data about them. That means users can explore connections they did not even set out to find. This ‘top-down’ approach is a drastic departure from the limited shared data users have today; sending spreadsheets through email and painstakingly combining them. They agreed it represents an exciting leap forward in generating new ideas.

• We first organized a class for ourselves on basic HIV science taught by local SMEs to ensure our later conversations with world-class researchers were constructive.
• We made multiple interactive exercises to understand scientists’ implicit needs and prioritize potential directions. We surveyed, interviewed, and observed over 20 diverse vaccine researchers, culminating in a document prioritizing the most valuable opportunities, establishing principles, and laying out the design challenges to be overcome.
• We made a Flash prototype that covered key scenarios. We tested iteratively-improved versions with users (RITE testing) before finalizing a design and detailed specification in PowerPoint. The tool was made in HTML, CSS, and JavaScript on top of the LabKey Server platform.
• After initial release, we collaboratively used the tool with researchers to understand successes and limitations, leading to a written evaluation of success.
• We created a product roadmap by identifying the achievable near-term features for a proof of concept and prioritizing future capabilities.

An excellent solution must help users understand data from other labs and use it properly. The Data Connector provides on-demand details about the experiments, data, and the labs that produced them along with contact information to encourage communication and collaboration on new ideas. Context is always available by hovering or in the side panel to prompt new ideas and caution users about potential complications.

We iterated with users to make sense of complex problems and refine a tool that isn’t just usable, but that will be used. The outcome: a flexible, interactive workspace that leads to new hypotheses, facilitates communication, and builds new connections among the scientists, advancing the cause of HIV prevention. Until now, there simply was no resource available to enable our best minds to work with all the pieces of the puzzle, and no one lab can discover a vaccine on their own.

Research discovered numerous counter-intuitive insights about successfully serving researchers. Excellence required us to take the harder path, following the evidence away from our initial assumptions. For example, the original vision called for rich online collaboration such as public comment threads attached to shared analyses. Our research showed that scientists, who value carefully constructed and peer-reviewed published arguments, would be repelled by such a tool. We focused instead on facilitating communication to promote scientifically valid ideas and increased offline collaboration.