I’ve been working as a digital designer at the Sabeti Lab of the Broad Institute of MIT and Harvard since June. There, I help design data visualization tools for genetic and clinical data, and digital products for both experts and non-expert users. A few months before joining the lab, I moved from Costa Rica to Boston to pursue a Masters in Information Design and Visualization at Northeastern University, where I’ve learned about data literacy, programming, statistics, systems thinking and design for experience. This has been an incredible personal and professional journey, and I want to share some of the things I’ve learned so far through a series of short stories. This first part will focus on why I ended up working in the field of genetic research and why I believe more designers should be interested in it.
I. Designers as mediators for discovery:
My first exposure to data visualization occurred several years ago while I was an undergrad in Costa Rica. I got an internship as a product designer at a local bioinformatics company and my eyes opened to a world of challenging information design problems. At the time my approach was inadequate at best: I couldn’t effectively explain to researchers why I was there to help them, or how the work of a designer was relevant to a biochemist or a cancer specialist. I think I was also a victim of the context, since trans-disciplinary collaboration in Costa Rica can be really hard to orchestrate. I did learn, however, that the world of bioinformatic tools was in desperate need of usability and HCI experts, and that whatever contribution I could offer would come in the way of a mediation for discovery: to provide an adequate space for exploration with better odds for fishing insights. This aspect of information design as a mediator for discovery is the foundation of my work:
as a designer my goal is to set the environment for the user and to offer solutions that can be embedded with as little pain as possible within existing workflows. It is not my goal to make scientific breakthroughs, only to discover and optimize the processes and flows that do.
However, before I talk about the combination of methods I’ve concocted to (try to) achieve just that, the question begs…
II. Why should digital designers care about bioinformatics?
Bioinformatics is a fast growing field and it’s the future of medicine and many other industries like agriculture and technology. It already is impacting the world in the form of novel treatments for diseases, and it’s making it’s way into a lot of us through ventures like 23&me. It’s expected that we will have sequenced up to 2 billion human genomes by 2025, surpassing the data size of Youtube and Twitter combined. What does this mean for bioinformatics? We may have the data, but we don’t have appropriate methods to visualize them, because even methods created two or three years ago have already become inadequate to manage the current complexity and size. While bioinformaticians are constantly struggling to find novel ways to analyze data, information and interaction designers have deep knowledge about visual representations of data, usability, accessibility and cognition, all of which are essential in highly interactive, exploratory data visualization tools. In other words, this is a major opportunity for designers to contribute to the next generation of scientific exploration.
There are many examples of low hanging fruit opportunities that we can tackle as designers: take the following visualization.
This is a visualization of protein mutation data from a portal of the the European Bioinformatics Institute called Ensembl. Regardless of what the dataset describes, this visualization could be redesigned to properly manage the complexity and density of the data. There are no true focus+context techniques, there is poor use of color palettes, and low visual accessibility and interactivity, not to mention that it looks quite dated. A couple of years ago I was part of a team that decided to improve this visualization:
This redesign is not a complete solution, but we believe it was an important improvement, and proof that collaborations between UX and UI designers and web developers and bioinformaticians can lead to great outcomes. So, back to the previous question…
III. How can designers work in bioinformatics?
I don’t have a straight answer (yet), nor will there ever be one for such a vague questions, but I hope that I come up with a collaboration framework (i.e. a series of generic methods and models) that can be replicated in different settings, whether it’s in academia or in the private sector. So far I’ve rounded up these premises and suggestions:
Transdisciplinary collaboration as culture
Scientists tend to be resistive to change, and unless there is an initial willingness to collaborate with other disciplines, there is little a designer can do within a research group or a company to improve their visual analytics or their digital products. If the company is going through a cultural shift and that’s the reason you got the job my advice is: try to gather as much information as you can and then visually demonstrate your abilities using inexpensive methods (wireframes, simple digital sketches, functional mockups with synthetic data), then present them to your stakeholders along with an overview of your process to start a conversation.
Pay attention to what they criticize the most about your ideas: this is where their research lies, and where you need to dig deeper. Also listen to their questions, especially if they sound something like “so in this map you show, could I see X by filtering Y across time?”. Your first sessions will be rough, almost awkward, but they’ll start to see the value you can bring to their work if you show them something they’ve never seen before (that makes sense to them).
Researchers are very logical beings (and this is a good thing)
It’s likely that you’ve had a PM or a client who simply can’t seem to understand how design process models work. Audit Frameworks, Behavioral Models like the User Journey and Convergent-Divergent maps are just that: models. These are nonlinear methods and tools with limited scope that help us do our work, but they’re not silver bullets, nor they are accurate representations of the real design process:
Scientists understand this very well because they use models all the time to manage and explain their own thinking, and in my experience it’s been far easier to explain my process to a scientist than to a PM, a developer, a client or an investor (no offense intended to any of you!).
Lay out your process, expose it and be ready to answer questions from your team, admit its shortcomings and modify accordingly.
Ignorance is a precursor for learning
“The acquisition of knowledge always involves the revelation of ignorance — almost is the revelation of ignorance”
– Wendell Berry
For a designer new to the field (and even for a veteran), it can be daunting at first to understand the complexity of the data, the goals of the scientists or even the subject matter itself. I spend a lot of my time making notes for concepts that I know I’ll have to read about later. At the lab we do “journal sessions” where researchers discuss and explain their published papers to a group of young students (and to me). This has been a very intimidating process, as I often feel like the most ignorant person in the room, but as per Berry’s quote, admitting ignorance is a precursor of learning.
Embrace ignorance, make notes of obscure concepts, take an online class on biochemistry, statistics or bioinformatics, and keep pushing forward. Only this way you’ll be able to gather enough knowledge to empathize with the researchers.
However, remember that ignorance is symmetrical, and that you’re an expert in usability, visual form, cognition and interaction (and they’re not). Many researchers will be interested in learning how to improve their figures and graphics for analysis and publishing. I’ve given small workshops on D3.js and short meetings on visualizing scientific data, heavily influenced by the work of Bang Wong and others.
Design as translation artifacts for shared understanding
Designers are exceptional at “translating” problems from an alien domain, to something they can systematically map out for shared understanding. In UX we sometimes refer to this as “alignment”, and use it to line up stakeholders, users, PMs and devs into the same values, goals and culture. (check out John Cutler’s articles for great thoughts on this!). We can do this too in bioinformatics if we’re open to learn about the work of our fellow researchers (see previous premise).
Some of us work on different projects simultaneously: one day I am conducting user research for a digital product regarding personal healthcare, and the next week I’m working on visualizing correlations between death rates and genetic mutations. The naivety that comes from the first exposures to each of these projects and their content is at the same time intense and an opportunity to think outside the box.
Stay naive and empathic. Keep mental models in mind but don’t be afraid to try completely new visual forms or other conventions if they provide better usability or accessibility. Use your sketches as a translational bridge between your vision and their needs.
These premises conform the foundation of my methodologies and approaches to designing for bioinformatics. In the next part, I will discuss specific (and probably more practical) strategies that any UX or Information designer can try in their own projects. These strategies are drawn from systems thinking, social psychology, information systems, and HCI.
I am creating this series to expose my design and thinking process, which can bring a lot of personal benefits, but I do hope that if you read this you’ll find it useful in your own work. Feel free to reach me on twitter, as I am looking forward to having discussions about optimizing these design processes and to learn about your issues, experiences, and ideas. Special thanks to Andrew Tang for his feedback on this article!