How do you track a lab sample from the vial to the result?

Forty vials of HEK293 go into a dewar on 17 September. Each one traces back to a single master vial frozen in January, and each came through one bad afternoon in Incubator 2.

A result is only as good as the link back to the material it came from. Which vial, how many thaws, did it pass mycoplasma, where has it been sitting? In most labs the person who did the work can answer, if they still work there. I'd rather the record answered. So I'll follow our example lab's new working cell bank from the master vial to the dewar, through the points where tracking usually breaks.

What does a sample record need?

A generated ID, what the material is, how much there is and in what form, its handling limits, its QC status and where it's stored.

• A generated ID. Never a typed description. "HEK293 WCB vial 3", typed three slightly different ways, is three samples as far as a search is concerned. The bank's vials are HEK293-WCB-P5-001 to 040. Their master vial is HEK293-MCB-P3-017.
• What it is. The sample type, and the cell line if it is one.
• How much, and in what form. Form is picked from a fixed list, and it decides whether an experiment takes a measured portion or the whole thing. A cryovial of cells is used whole. A bottle of conditioned medium is drawn from. Get the form wrong on day one and every stock figure after it is wrong too.
• Handling limits. Freeze and thaw cycles, single use once opened, shelf life.
• QC, and where it is, down to the shelf.

In CCA Labs the sample page opens with four cards: the amount available, the storage temperature, QC and the days until it expires.

Why is QC two separate things?

Because the assays are evidence and the release is a decision. The bank's mycoplasma PCR comes back negative on 24 September. Someone then releases the bank for use. CCA Labs keeps the two as separate controls, so a list of passing assays doesn't release a sample on its own.

The bank was grown and frozen without antibiotics, which is common practice before a mycoplasma test. Antibiotics can hold a low-level contamination below what the test will pick up, without clearing it. Assays can also be entered when a sample is created, for material that arrives with its results already in hand.

How should samples be labelled?

With a code that scans. Each sample in CCA Labs has a QR code that opens it at the bench, from a phone camera in the browser. You print a label from the sample, tube labels for its aliquots, or a batch of 40 from the registry for a bank like this one.

The same QR scheme covers inventory lots, equipment, storage locations, workflows and experiments, so the scanner opens whatever it's pointed at. Each scan is recorded against the sample, which helps later when you're working out who handled what.

Where has it been?

Record the location as a path: the lab, the equipment, the shelf. The bank sits in Dewar 1, rack C, box 4. In CCA Labs the location is pinned to equipment in the Virtual Lab, so Open in Virtual Lab goes to the exact shelf, and What else is in here? lists its neighbours.

Where the material has been matters as much as where it is. On 15 September at 14:06, CO₂ in Incubator 2 falls to 4.2%, under its 4.5% alert threshold. The facility manager acknowledges it at 14:21: the supply had switched to cylinder B. The eight P5 flasks the bank came from were on that shelf.

When a sensor such as a Pebble watches the equipment, the sample page shows its live readings and history against the acceptable range. So "was it ever out of range?" has an answer instead of a shrug. A quarter of an hour at 4.2% is probably harmless to HEK293. The point is that nobody has to guess.

Splitting into aliquots

Split into Aliquots asks for the number of tubes and the volume in each. Aliquots are tubes of the same sample, not new samples: they share its chain of custody, and each is available, in use, consumed or discarded. Up to 96 tubes can be edited at once, which is a plate or a full freezer box.

What happens when an experiment uses it?

The use is recorded on both sides, so the sample shows where it went and the experiment shows what it used.

Before material is used, CCA Labs checks its handling limits. If a vial has reached its freeze and thaw limit, or it's single use and already opened, the scientist gets a warning that says so. It's a warning, not a block. A hard block tends to get worked around by copying the record, and the copy has lost the history the block was protecting. So the scientist decides, and their acknowledgement is saved with the use.

Who has had it?

Chain of custody is the ordered list of who held the material and when it changed hands. In CCA Labs it's one list per sample, oldest first: received, transferred, returned, destroyed, placed in storage, removed from storage. Record Custody Event takes the receiving lab, the courier or collaborator, and a note for the next person.

The parties are names, not user accounts, because one side of a handover is often outside your lab. Entries are added and never edited.

What if the record is wrong?

Record a correction with the reason, and never edit the value in place. A silent edit is the worst fix, because everything that already used the sample now points at a record that says something different from what was true at the time.

• Correct identity is for a cell line or passage recorded wrongly, say the pool entered as P4 instead of P5. Both are fixed once set, and this is the only way to change them. It asks why and keeps the earlier value.
• Correct the form is for material recorded as the wrong kind of thing. Because form decides how experiments draw on it, this creates a replacement record with the right form and links the two. The original stays as it was.

How does a sample leave the lab?

In one of four ways, and the record should say which. Consume material records a loss outside an experiment, with a reason. Return or destroy asks who authorised it and why. Retire takes it out of use but keeps it, restorable at any time. Move to bin is for records made by mistake, restorable for 30 days and then permanently deleted.

Following a result back

Once every step is an event, lineage follows from the record. When the bank's freeze step produces the 40 vials, each records what went into it: the pool it came from, its line and its passage. The pool records its flasks, and so on up. From any vial, CCA Labs traces parents and children across samples and cell lines, back through the pool and the eight flasks to master vial 017. Access is checked at every step, and when part of the tree is hidden from you, the view says so.

A freezer spreadsheet can't do this, as why the freezer spreadsheet always drifts explains. CCA Labs has limits of its own: it isn't built for patient samples or biobank consent, and it doesn't drive your instruments. When you create your lab, the AI drafts your workflows and inventory from the protocols and spreadsheets you upload, so the first samples you add have experiments and storage to link to. Cell line identity testing and the rest are on the all-in-one lab page.

The last number in this story isn't on a sample page. The bank cost £510.39 in materials and bench time, £12.76 a vial, and that figure shows on the project the lab lead funded with £2,500 on 1 September. That's where the story started, in how to track a lab budget across grants and projects.