Why Every New Customer Starts on VarSeq 3
Today, every new Golden Helix customer starts on VarSeq 3, our clinical variant analysis platform, and for good reason: sharper structural variant handling, a rebuilt assessment catalog system built to scale with years of clinical data, refined annotation, and a stronger reporting engine. Some of our longest-standing customers held onto VarSeq 2 well after VarSeq 3 shipped, and that is a compliment to VarSeq 2, not a knock on the new release. Years spent tuning a workflow around a lab’s exact clinical process is not something anyone abandons casually, and we have always thought customers should move on their own timeline. Most do move eventually, for the same handful of reasons: better structural variant handling, catalogs that scale further, algorithms that have simply moved forward. The upgrade is worth making, and the last few months gave us two good examples of what is worth understanding about VarSeq 3’s features before you trust the new numbers they produce.
What to Check First
VarSeq 3 categorizes structural variation more granularly than VarSeq 2 did. In VarSeq 2, indels and copy number calls mostly stayed within their own tables, and the break-end table was reserved for true rearrangements like translocations and inversions. VarSeq 3 draws those lines more precisely, which allows for more customizable and robust CNV reporting. The tradeoff is that some larger indels and copy number calls can now show up in more than one table at once, not because they are new findings, but because VarSeq 3 can now show an angle VarSeq 2 never could.
One lab saw this firsthand after migrating a project: its structural variant and breakend tables grew from 1,300 entries to 11,000 records. The number looked big, but the explanation was straightforward once the lab compared it table by table against its VarSeq 2 results rather than judging the total on its own. Once a single filter was added to keep the break-end table to true rearrangements only, the count landed right back in line with what the lab expected to see.
That is the habit worth building whenever a number changes after an upgrade: look at what is behind it before deciding whether it means anything. In a system built to say more about your data, not less, a bigger number is often a better view, not a bigger problem.
What to Check Next
VarSeq 3’s catalog system is built to automatically capture CNV calls, gain, loss, or normal, as a lab reviews them, so that classification becomes part of the lab’s institutional knowledge instead of something that has to be re-decided every time a familiar variant comes up again. That automatic capture is the whole point of an assessment catalog: years of expert judgment, ready the next time it is needed.
A different lab put this to the test while migrating years of copy number classifications into VarSeq 3’s updated catalog system. The migration itself is exactly this capability in action, carrying forward gain, loss, and normal calls a lab has already made. Because a catalog is only as useful as what is actually captured inside it, it is worth confirming after any migration that classifications came through the way you expect, a handful of known gains, losses, and normals checked against what your team already knows to be true.
That is the second habit: understand what a feature is supposed to be doing, and confirm it did it, the same way you would check a new report template against a known case before relying on it.
Why This Is Part of Our Job, Not Just Yours
Both of these moments point at the same thing. An upgrade that runs cleanly is not the same as one that behaved exactly as expected, and closing that gap starts with understanding what a feature is supposed to do and checking that it did it, rather than assuming either version is automatically right. That is why Golden Helix builds features like assessment catalogs in the first place, so a lab’s accumulated variant knowledge moves forward across versions instead of resetting, and why our team treats “the numbers look different than I expected” as a real question worth a real answer. Precision medicine only holds up if the data behind it stays trustworthy at every step, including the step where you upgrade.
Before You Upgrade
If you are moving from VarSeq 2 to VarSeq 3, spot-check a few familiar results against the new version before fully trusting it: a count from a structural variant table, a handful of entries from an existing catalog. And if something still does not add up, reach out. We are glad to help you understand what you are seeing.
