patric
What is PATRIC?
PATRIC is an online resource of bacterial and viral bioinformatics with a biological hierarchy breakdown from taxonomy to genomic features. The website supports biomedical community to research on infectious disease and potential pathogens based on genomic comparisons and comparability. Users were also able to annotate their own genomes in the database, where about 10,000 + genomes were annotated every year.
Why the change?
In 2015, first attempt to rebuild the infrastructure of the past PATRIC website was set forth in order to move from a serial to parallel back-end architecture to scale. Accordingly, front-end engineering was also debated, where UX team (my professor and I) joined the effort to redesign the areas of change from back-end infrastructure and to provide efficient ways of search, analyze, and compare genomes from the table of vast bioinformatics data.
Team
Project Manager
Domain Expert
Domain Expert
Domain Expert
Developer
Lead UX Researcher
UX Assistant
Phase of the Project
Research
Analyze
Conceptualize
Prototype
User Test
Iterate
The front-end design efforts have already been set in place when I entered the scene in May 2016. The instructions given to me were: “Do an evaluation of the alpha site with an outsider’s perspective without any background in bioinformatics.”
My role...
Walkthrough
I evaluated the website, strictly on UX/UI standards and provided a list of action items to update the website. The presentation was made via visual samples and doing a walkthrough of each main tabs in the website. Below are some of the sample suggestions made to the team after discussion with the lead UX researcher.
Disable inaccessible tabs
Disable inaccessible action icons
User Test
We conducted a summative evaluation with the 4 primary users, 1 from the design team and 3 outside of the team. We came up with four benchmark tasks, coupled with exploratory tasks and probing questions as to what the user expects to see before clicking an icon or a tab. Interesting behaviors were further probed on the spot, such as extreme frustration or confusion with current state within the site.
The participants were asked to talk aloud throughout the task, they were also inquired to provide any emotional feedback. Each test session lasted between 1 hr - 1.5 hrs. Only qualitative data were taken through extensive note taking by 2 team members (a project manager and me) and my professor moderated the session. Critical incidents were also noted to probe during the debriefing session.
The development and project management team also tested the design at the Chicago national research center, HQ.
4 pages of benchmark tasks and questions
Iterate
The qualitative data were filtered for unnecessary comments and the two notes were compared to fill-in incomplete notes and confirm important points. The consolidated data were categorized into 17 categories.
A checklist table was built, providing the UX issues, the category of issue, and potential design implications and solutions. The checklist was agreed to be an efficient way to visualize and organize the data and design solutions, also to keep the site changes transparent with the entire team, which issue was resolved or ruled out for other technical limitations or stakeholders’ directions.
Challenge
One thing we had to clarify was terminology. Apparently, even in the bioinformatic field, people were using different terms for different reasons and definition. Moreover, there were myriads of abbreviations thrown around without a glossary, which forced me to keep going back to the past website’s help page for definition. For example,
Perspective vs View vs Model
Features vs Genes
Transcriptomics vs Scholarly Literature
This had a huge impact on:
1. Search navigation path for the users, because some users won’t be expecting to see genes under features, and would most likely ignore the tab that is not familiar to them
​
2. Design of the icons. Most icons were made in-house and linked to no mental model of the users
"Speak user’s language"