Keep the Prior Protocol Id After You Publish the Next

An implementation checklist for versioning laboratory protocols is a protocol-ID walk, not a new filename. A new PDF is not a version until the old ID still resolves. Prefer publishing the next method only when a dated run can still open the prior file. Reject final_v8.pdf that hides v7.

Protocol deviations still own the unfollowed step. Plasmid versions still own parent and diff on a map. Protocol versus experiment record still owns the recipe versus the run. This page only versions the recipe. It does not record a deviation. It does not version a plasmid. It does not decide whether an SOP name is required (that naming page was not opened in this batch).

A New File Is Not a Version by Itself

Implementation checklist for versioning laboratory protocols is the gate a lab manager runs when a method changes. The outcome is not a newer PDF. The outcome is two resolvable IDs: the method that will be used next, and the method that was used on last Tuesday. If the old ID dies when the new file is saved, historical runs lose their recipe. The lab has a replacement. It does not have a version.

The search asks how to version laboratory protocols. Drop the product brochure. The walk below is an operational heuristic. It is not a document-control statute and not a software feature list. A manager who only asks “did we upload the new file?” will orphan every experiment that cited the old name.

Walk the Protocol-ID Checks

Run the walk on the method object the lab will cite, not on a slide deck that happens to describe the method.

1. Freeze the current ID. Write the identifier last week’s runs used, the file that ID opens, and the date that file was last authoritative. Checkpoint: a second person opens that same file from the ID alone.
2. State the change. Write what will differ in the next method: a reagent, a time, a temperature, a step order. Checkpoint: the change is specific. “Updated protocol” is not a change statement.
3. Mint the next ID before anyone replaces the file. Do not reuse the old token. Checkpoint: both IDs exist as strings before the new file is published.
4. Publish the new file under the new ID. Point the lab’s current-method slot at the new ID. Checkpoint: today’s empty record would cite the new ID, not a nickname.
5. Keep the old file findable. The old ID must still open the old file. Checkpoint: last Tuesday’s record still resolves.
6. Retire, do not delete, unless policy truly requires destruction. If destruction is required, write that the ID now resolves to a tombstone that names the destruction date. Checkpoint: the ID does not 404 into silence.

If any checkpoint fails, stop. Do not overwrite. Overwrite is how v7 becomes a rumor. Plasmid version operations use parent and diff because a map is a sequence object. A protocol is a method object. Do not copy that walk here except for the shared idea that a new filename is not a version.

A worked fail is a shared drive file named PCR_master.docx that is edited in place every month. Last month’s annealing time is gone. Experiments from March still say “PCR_master.” Intake of a new ID never happened. Versioning failed at the freeze checkpoint. The fix is PCR_master_2026-03 still opening, and PCR_master_2026-04 becoming current. A second fail is a PDF emailed as “final_v8” while v7 is deleted “to avoid confusion.” Confusion is the point of a version system. Deletion created it.

Expected Result and Verification

The walk is done when a dated run can name the old ID and open that file, and a new run names the new ID. Verification is not a newer timestamp on one document.

• The prior ID still opens the prior method.
• The current-method slot points at the new ID.
• The change statement is written.
• A run from last week does not silently inherit the new steps.

Common failures are a “current” shortcut that was retargeted without a new ID, a PDF that replaced the only copy, and a deviation written because nobody could find the old method. A deviation is the wrong object for a planned method change. If the step was supposed to change for everyone, version the protocol. If one run departed from the still-current method, use the deviation sibling. If verification cannot be completed, you published a file. You did not version a method.

Hold the IDs After They Exist

Use a record surface after the IDs exist, not as proof that they exist. After the walk, a workspace such as ZettaNote can hold dated protocol objects and the experiment records that cite them. Official product language includes templates, procedures, and version-related history. That is a place the two IDs may live. It is not a version certificate. Software alone does not complete 21 CFR Part 11. A versioned protocol is not a compliance event. Golden Gate is not a Zetta feature and does not belong on a method-ID walk.

If the lab still edits one living document and hopes memory will sort last month from this month, you do not have versioning. You have a file. Keep the prior ID. Then publish the next.

Experiment records break first when the protocol ID is a nickname. “Use the PCR protocol” is not an identifier. Last Tuesday’s run and next Monday’s run will both write that sentence and mean different files. The close-out sibling already asks for a protocol version on each entry. This walk is what makes that field resolvable. If the entry cites PCR_master_2026-03, someone must still be able to open PCR_master_2026-03 after 2026-04 is current. If they cannot, the entry is complete in form and empty in method.

Shared drives hide a second fail: shortcuts named current that are retargeted in place. People bookmark current and never notice the ID changed. Treat current as a pointer that must be updated in a recorded step, not as the identity of the method. The identity is the durable ID. The pointer is a convenience. If a convenience overwrites history, retire the convenience. Paper binders fail the same way when the latest printout is stapled over the old one and the old one is discarded. The staple is not a version. The discarded printout is the missing ID.

Frequently Asked Questions

Does saving a new PDF create a protocol version?

No. The prior ID must still resolve. A new file that hides the old method is a replacement, not a version.

Is a protocol version the same object as a protocol deviation?

No. A version replaces a method. A deviation records an unfollowed step on the method that was still current.