User-Centered Design in Bioinformatics
-
Ethan Cerami
Usability is an often overlooked feature in bioinformatics software. The result is that many bioinformatics tools remain underutilized, and many fail to reach their full potential.
The EMBL European Bioinformatics Institute (EBI) is one of the few bioinformatics groups to specifically embrace usability and user-centered design in bioinformatics (see recommended reading). Many of the group’s recommendations are compelling, but it’s hard to imagine most bioinformatics groups following the rigorous example put forth by the EBI — for example, putting in the overhead to define user “personae”, conducting extensive user interviews, and setting up off-site workshops with potential users to define workflows.
Nonetheless, there are three areas where the EBI is spot on.
First, you must champion the idea of usability yourself. For most groups, the end-game is the paper describing the software, not the software itself. As such, there can be very little incentive to create user-centered bioinformatics software in the first place. This is changing for major bioinformatics resources, where funders consider indirect measures of usability, such as downloads, web hits and scientific citations. But, the bottom line is that you must first convince yourself and your team that usability is a worthy undertaking. Start by asking yourself who your intended users really are, whether you want to make the effort to truly connect with them, and rally your group to adopt one or more usability practices.
Second, before building any kind of user interface, consider building paper mockups and paper prototypes. These can be a cheap, effective tool for building more usable interfaces, and getting your entire team on the same page. They can also save a huge amount of time, because it’s easy to modify a protoype — and, not so easy to modify real, working code. The recommended reading outlines specific practices regarding paper prototypes, but the best advice is to start with a good tool (I highly recommend Balsamiq), create a few mock interfaces, and start asking colleagues for feedback.
Third, if you really want people to use your software productively, usability testing is the single most important new practice you can adopt. Plus, it’s surprisingly easy and fun, and you need not be a usability expert or even consider hiring one. When I was previously involved in cBio Cancer Genomics Portal, we purchased Rocket Surgery Made Easy, followed the exact steps outlined in the book, and performed several rounds of usability testing with graduate students, post docs, and staff. The process was simple, and succeeded in uncovering multiple usability problems that we were simply unaware of.
Importantly, you don’t have to follow every step employed by the EBI. Rather, you can start by focusing on just these three areas, and iterate. With any luck, this minimalist approach could be adopted by just about any bioinformatics group, and your users and the scientific community at large will thank you.