Why Two Laboratories Get Different Results From the Same Peptide
Almost everything that decides the outcome of an assay is settled before the plate goes in the reader, and by default almost none of it gets written down. That single fact explains most of what looks like irreproducibility. When two competent groups run what they take to be the same experiment on the same compound and land on different numbers, the reflex explanations are operator error or bad material. Either is possible. Neither is likely, because the list of things that had to match and were never specified is longer than any methods section admits.
Closing the gap: what actually helps
Since the diagnosis takes a while, here is the prescription first.
- Specify material completely: supplier, lot, stated purity, net content and salt form, not merely a compound name.
- Measure concentration instead of assuming it, using absorbance wherever the sequence permits, as in concentration by A280.
- Hold a single lot across an entire study where that is feasible, so lot variation cannot disguise itself as a result.
- Document the whole preparation chain rather than only the final concentration.
- Include a positive control with a known response. A control behaving as expected is what distinguishes a genuine negative from an apparatus that simply was not working.
Three words, three different problems
One term gets used for three separate questions, and the fixes are not interchangeable.
| Repeatability | One laboratory, one operator, one set of instruments, run a second time. When this fails, the method itself is at fault, and tightening the protocol is the remedy. |
| Reproducibility | A different laboratory, different hands, different equipment, working from what was published. Failure here almost always means something went unrecorded, and the remedy is writing down whatever nobody wrote down. |
| Replication | Another group pursuing the same biological question, not necessarily by the same route. Of the three, failure at this level may carry the most information. |
Everything that happens between vial and well
A compound reaching an assay has already been weighed, dissolved, aliquoted, frozen, thawed and diluted. Each of those operations is somewhere two laboratories quietly part company without either one logging a deviation.
Weighing brings the static and hygroscopicity difficulties described in weighing lyophilized peptides. Volume transfers bring systematic error, covered in pipetting accuracy; an instrument out of calibration by three percent low is both invisible and perfectly consistent. Dilute working solutions shed material onto surfaces, as in adsorptive loss, and how much is lost depends on whatever labware happens to be in each building. Storage diverges too, including the unplanned cycling set out in the auto-defrost problem. None of this reaches print. All of it alters how much intact compound is really there.
Why the certificate gets blamed first
Two lots of a peptide are not the same thing, and several of the differences stay hidden unless somebody goes looking. Net peptide content varies between lots, which means equal weighed masses are not equal molar amounts; see net peptide content. Counter-ion can differ as well, shifting the mass fraction and altering what else lands in the well, per counter-ions and salt form.
Purity numbers from two vendors need not be comparable even when both are reported in good faith, for reasons laid out in why certificates disagree on purity. Where an impurity happens to be biologically active, lots carrying identical stated purity can still behave unlike each other because their impurity profiles differ.
All real sources of divergence, and all of them get accused early, largely because a certificate is the only document in the chain and therefore the only thing available to accuse. The bigger differences usually sit downstream.
The variables nobody wrote in the methods
Contribution from the biological model is at least as large. Passage number, receptor expression, serum lot, confluence and authentication status differ from building to building and are rarely reported in full, which is the ground covered in cell line choice.
Serum earns a paragraph of its own, being the least controlled reagent most laboratories use. It is a biological product whose composition, peptidase activity and growth factor content shift from lot to lot. Switch serum lot partway through a study and a variable has changed that nobody is tracking.
Disagreement is not misconduct
Against that inventory, two capable groups reaching different conclusions is the expected result. The compound was nominally identical; exposure, model and measurement were each a little different, and effects in this literature are frequently modest enough that those small differences are the same magnitude as the effect itself.
A publication asymmetry compounds it. Experiments that work get written up, experiments that do not usually go unpublished, so the visible record over-represents the conditions under which something was seen. One positive report is weaker evidence than it appears, which forms part of the reading discipline in what the published literature actually shows.
Where a vendor’s contribution ends
Traceability on the material side is something a supplier can deliver: lot identification applied consistently, per batch and lot numbering, a certificate that reports values instead of verdicts, and an unambiguous statement of what a quoted mass refers to.
The remainder of the chain is beyond any vendor’s reach. Handling, model and measurement get assembled locally, and they account for the larger share of the variance. An honest account of a reproducibility failure starts at the vial and seldom finishes there.
