Designing clarity into complex genomic workflows
The starting point
SeqSMART is a clinical genomics platform where specialists analyze genetic variants, review scientific evidence, and work with large amounts of complex genomic data.
The platform already had the data and functionality its users needed. The problem wasn’t what the product could do — it was how that information was presented.
Tables were crowded, visual hierarchy was limited, and large amounts of complex information were presented without enough structure to make them easy to scan and understand.
So the redesign wasn’t about adding more features or removing scientific detail. Genetics professionals rely on that information. My role was to find a better way to organize and present it.
The information was there — the design just wasn’t helping it speak.
That led to the main question behind the redesign:
How do you make a complex scientific workflow feel easier without oversimplifying the science?
ACMG Classification
ACMG Classification was one of the most complex parts of the platform.
Specialists work through multiple categories of scientific evidence, each containing different criteria, evidence strengths, supporting information, comments, and classification states. Presenting all of this as one long list would make an already complex process even harder to navigate.
The criteria were already organized into eight evidence categories. I redesigned how these categories and their criteria were presented, giving specialists a clearer overview of the classification before they move into the details.
Each category shows the current state of its criteria — Met, Unmet, and Unassessed — while selecting a category reveals its detailed evidence without losing the context of the overall classification.
See the whole picture. Focus on one decision at a time.
Judging, not just accepting
The system provides a result based on the available evidence, but the specialist still needs to be able to review that result and make their own judgment.
When a specialist changes a system result, the previous result shouldn’t simply disappear.
I designed the override state to preserve the context around that change. The interface shows who changed the result and when, keeps the original system result accessible through View System Result, and provides Restore Previous when an earlier state needs to be recovered. This keeps the specialist in control while making changes visible and traceable.
The specialist makes the decision. The interface keeps the context.
Designing for multiple doctors
The classification workflow also needed to support more than one doctor reviewing the same case. That meant the interface needed to make it clear who made a change, what changed, and how discussions around the evidence relate to those decisions.
Changes are attributed to the doctor who made them, comments stay connected to the relevant evidence, and the current decision remains distinguishable from the discussion around it. Managers can also review changes and restore a previous state when necessary.
Living with dense data
Variant tables presented a different design challenge. Each case can contain a large amount of information, with multiple attributes competing for limited space. Putting everything directly into a single row would make the table difficult to scan.
I used information hierarchy and progressive disclosure to separate what users need for quick scanning from what they need during deeper investigation. The most important information remains visible in the table, while additional details can be accessed when needed without losing the user’s position in the data.
Sanger Analysis
Sanger Analysis required a different kind of experience. Here, users move between uploaded AB1 files, detected variants, chromatograms, genomic information, and different representations of the same genetic data.
Instead of treating each of these tools as a separate destination, I brought the related information together into one continuous analysis workflow. This makes it easier to move from raw data to the information needed for further analysis without repeatedly switching between different parts of the platform.
What changed
No unnecessary new features were added. The value of the redesign came from changing how the existing information was structured and experienced.
From scattered to structured. From overwhelming to scannable. From unclear states to traceable decisions.
Rather than hiding scientific detail, I used visual hierarchy, progressive disclosure, consistent interaction patterns, and clear grouping to make that detail easier to work with.
Complex products don’t always need less information. They need better structure.