KBase

KBase
Client: Truss, KBase
Dates: June 2020 – January 2021
Skills/Subjects: , , , , , ,
URL: https://www.kbase.us

I led research for and co-designed critical new functionality to the Department of Energy’s Systems Biology Knowledgebase (KBase), allowing biologists and bioinformaticians in national labs, universities, and companies to better upload and analyze data for genomics and bio-energy research, especially for metagenome analysis.

Truss was hired primarily to improve product management, design, and agile software development best practices among KBase’s engineers, scientists, and educators. So first we sought to understand how biologists and bioinformaticians use KBase and what their motivations are. We conducted interviews and meetings with KBase personnel to understand the organization, how people work together, and the vision of the product. Finally, we presented our recommendations on process improvement and how to apply them to the process of building a feature.

Screenshots of slides displaying process, project, and role maps

Like much of the work I did at Truss, my KBase work was simultaneous process coaching and demonstrating that process through actual product development. To do this, we formed a blended contractor/client end-to-end delivery team, resulting in:

  • Reduced siloing, encouraging knowledge sharing
  • Increased quality and speed of delivery
  • Improved ownership of services and systems 
  • Decreased organizational complexity and bureaucracy
  • Made team formation flexible where possible
  • Reduced meeting overhead 

Through our teamwork, our co-created feature was an improved workflow for users to upload files and a more stable and broadly usable batch analysis tool. We also created an initial digital design system for both the Joint Genomics Institute (JGI) and KBase, based on their existing print style guides and the US Web Design System. This was taken further in subsequent Truss contracts with JGI.

Data Panel and Import to Staging Area

I focused primarily on generative and evaluative research, while my design partner Mallory Robertson focused on UI design. We both traded off and complemented each other on tasks, so those weren’t exclusive roles, while working very closely and cheerfully (honestly, best team dynamic ever) with our product and engineering partners on both Truss and KBase sides.

Through participatory and other design research activities, we synthesized our blended teams’ and users’ goals, pains, and ideas over a couple months, and we worked with our PMs to set them as clear project objectives. Primarily that users can easily bulk upload certain data types, convert them to other types, and then perform meaningful analyses to generate high-value products.

Screenshot of a whiteboarding activity showing "How might we" statements

Through an iterative design-build-test process, we naturally ran across many usability issues along the way, at least providing clearly scoped tickets if not fixing them during our project as time and scope allowed. Our final outcomes were:

  • Delivery of this high-value data upload feature 
  • A big step towards proving that more datasets in the system helps users get meaningful results (an external, but major, client business goal)
  • A prioritized roadmap, including those usability tickets
  • More streamlined processes as an organization
  • More collaborative team dynamic