The place lab software breaks is the handoff

by Tom BergLab Systems Editor, HMND

The failure usually shows up in a boring room at 4:12 on a Tuesday, when a sample has a barcode in the LIMS, a note in the ELN, raw files in the SDMS, and one instrument that still spits out a cursed CSV because nobody had the patience to bully the vendor into a cleaner export. The science keeps moving, the people keep moving, and the system quietly asks someone to copy the truth by hand.

The gap between systems is where provenance goes soft

Lab informatics vendors like to talk about the suite as if the suite is the story. The story is the seam.

When labs run separate ELN, LIMS, and SDMS tools, or keep paper and point systems around the edges, they create silos and manual copying between systems, which is where version drift and validation trouble start. That is also where sample lineage gets muddy, because the ELN holds the reasoning, the LIMS holds sample state and release status, and the SDMS holds the raw instrument data, but none of them can rescue a handoff that was never modeled properly.

You can see the mess in the little things. A hidden character in a file name breaks an interface. A vendor API changes. The middleware passes the happy path in test and coughs in production when the edge case shows up. Then someone in QC opens Excel, because the batch still needs a number and the system is busy pretending this is an exception instead of the daily weather.

In pharma, that kind of drift matters because software failures can affect product quality and patient safety, and traceability is a real regulatory demand rather than a nice dashboard label. A bad handoff can distort provenance, delay release, or leave a reviewer trying to reconstruct a sample’s path from three systems and a memory that is already tired.

Migration pain is mostly human pain with a technical receipt

I have watched teams spend months on integration design and still miss the step where the analyst swaps one instrument for another because the first one was under maintenance and the backup exports a different header set. That is not drama. That is Tuesday.

The hard part is that ELNs, LIMS, and SDMS tools were built for different jobs. ELNs document experiments, LIMS track samples and workflows, and SDMS archives raw instrument data. In practice, many labs still route coordination through email, spreadsheets, and a few stubborn people who know where the bodies are buried. Marina would call that culture. I call it a shadow system with better uptime than the official one.

Release gets delayed when the chain between instrument, storage, and sample record cracks. Traceability gets weak when a sample result lands in one place under one identifier and is later retyped into another place under a slightly different one. The lab still gets work done, but the audit trail starts looking like a bar napkin after closing time.

A migration plan that ignores instrument reality is a wish. A migration plan that ignores the workaround the bench already depends on is worse, because it will fail after go live, when the people who knew the tricks have stopped answering questions. If your stack cannot survive a messy handoff, it looks sturdy only from the procurement slide deck.

This is why integration pressure keeps climbing while labs talk about modernization with a straight face. The real test is whether the system can absorb bad exports, awkward naming, maintenance swaps, and one exhausted scientist who just wants the batch to release before lunch. Daniel would call it an API problem. The bench calls it the reason they still keep a local spreadsheet.

Labs do not need more glossy diagrams of connected platforms. They need the ugly middle to be reliable, the CSV to stop being cursed, and the human detour to be visible before it hardens into policy by habit.

If this handoff problem is sitting on your desk, write to hello@example.com. We build the systems side of that mess.