How to Read a Peptide Certificate of Analysis

A certificate of analysis, usually shortened to COA, is a record of measurements taken on one specific production lot of material. It is issued by whichever laboratory ran the tests, and it reports what the instruments found on the date the sample was pulled. It is a data document. It is not a marketing claim, not a guarantee of performance, and not a statement about safety.

A useful COA answers two separate questions. First, is the molecule in the vial the molecule named on the label. That is an identity question, and it is answered by mass spectrometry. Second, what fraction of the material in the vial is that molecule rather than something else. That is a purity question, and it is answered by chromatography. Documents that answer only one of the two are incomplete, and documents that answer neither with primary data are decorative.

Most of the value in a COA sits in the attached instrument output rather than the summary table on the front page. A number typed into a table can be typed by anyone. A chromatogram with labeled axes, retention times, and an integration table is a piece of evidence that can be checked. Learn to read the attachments and the front page becomes far less interesting.

Start at the top of the document. A complete header identifies the material and the lot with enough precision that the document cannot be quietly reused. Look for the product name, the full amino acid sequence in one-letter or three-letter code, the molecular formula, the theoretical molecular weight, a lot or batch number, the manufacture date, the analysis date, the appearance, and the storage condition the material was held at.

The sequence and the formula should agree with each other. If a document lists a fifteen residue sequence and a molecular weight that would be plausible for a forty residue chain, someone assembled the paperwork carelessly at best. Peptide molecular weights are trivially calculable from sequence, and a mismatch between stated sequence, stated formula, and stated mass is the single fastest way to detect a fabricated document.

Note also whether the document distinguishes average mass from monoisotopic mass. On a peptide of two thousand daltons the two differ by roughly one to two daltons, which is enough to look like an error if you compare the wrong pair.

Peptide purity is almost always measured by reverse-phase high performance liquid chromatography. The sample is dissolved and injected onto a column packed with silica particles bonded with C18 alkyl chains. A mobile phase gradient, typically water and acetonitrile with about 0.1 percent trifluoroacetic acid as an ion-pairing modifier, is pushed through the column at increasing organic strength. More hydrophobic species stick to the C18 phase longer and elute later. A UV detector at 214 nm watches the column outlet, because the peptide bond itself absorbs strongly at that wavelength.

The resulting plot has retention time in minutes along the x axis and absorbance in milli-absorbance units along the y axis. The tall peak is your peptide. Everything else is either an impurity, a solvent artifact, or noise. A well run separation shows a flat baseline sitting near zero, a sharp and roughly symmetrical main peak, and clear vertical drop lines where the integration software decided each peak began and ended.

Read the shape of the main peak, not only its height. Fronting or heavy tailing suggests column overload or a degraded column, and either compromises the integration underneath. A shoulder matters more than a small separate peak elsewhere, because closely eluting species are usually closely related species: a deamidated variant, a diastereomer from partial racemization, or an oxidized form. Those are the impurities hardest to separate and easiest to hide.

Check the early region too. Salts, residual scavengers, and injection solvent come through near the void volume in the first minute or two. Honest methods exclude that solvent front from the integration and say so.

Chromatography tells you how many things are present. It does not tell you what they are. Mass spectrometry supplies the identity half of the answer by measuring the mass to charge ratio of ionized molecules.

For peptides below roughly ten kilodaltons, electrospray ionization, or ESI-MS, is the usual technique. It puts multiple protons on each molecule, so the raw spectrum shows the same species at several charge states, labeled [M+2H]2+, [M+3H]3+ and so on, which software deconvolutes into a single neutral mass. MALDI-TOF is the alternative, more common for larger peptides, and it generally produces a singly protonated [M+H]+ ion sitting about one dalton above the neutral average mass.

Compare observed against theoretical on the same basis. A standard quadrupole ESI instrument reporting average mass is normally within about one dalton. A higher resolution instrument reporting monoisotopic mass should agree within a few parts per million. A mass off by twenty or thirty daltons with no explanation is a different molecule, not a rounding error.

Learn a handful of diagnostic mass differences and the trace starts talking. A satellite peak at plus sixteen daltons is oxidation, most often at methionine. Minus seventeen suggests loss of ammonia or pyroglutamate formation from an N-terminal glutamine. Minus eighteen is dehydration. Plus forty-two is acetylation, and can also arise from incomplete capping chemistry. A peak lower than the target by exactly the residue mass of one amino acid is a deletion sequence, meaning a coupling step failed during synthesis. A peak at roughly double the target mass is a dimer, frequently disulfide-linked when cysteine is present.

The purity figure on a peptide COA is almost always chromatographic purity by area normalization. The software integrates the area under every peak in the chromatogram, sums them, and divides the main peak area by that total. Multiply by one hundred and you have the number printed on the label.

That calculation carries assumptions. It assumes every species present absorbs at 214 nm in proportion to its quantity, which is approximately true for peptidic impurities because they all contain amide bonds and badly untrue for anything that does not. It assumes everything in the vial elutes during the run. And it assumes the integration boundaries were drawn honestly.

So chromatographic purity is a ratio among the things the detector could see, not a statement about the mass in the vial. Trifluoroacetate counterions, residual water, and inorganic salts have no meaningful absorbance at 214 nm and no C18 retention, so they never enter the denominator. A lot at 99 percent chromatographic purity can still be well under 99 percent peptide by weight. That is not deception, it is what the method measures.

A COA is a snapshot of one sample at one moment. It says nothing about what happened to the material afterward. Storage temperature during warehousing, transit time, and thermal excursions in shipping are all outside the document. A lot that tested at 99 percent in March is not certified to be 99 percent in November.

Unless separate tests are listed, the document is also silent on sterility, bioburden, and bacterial endotoxin, and equally silent on water content, counterion content, and residual solvents. Each requires a dedicated method: a compendial sterility test, a limulus amebocyte lysate or recombinant factor C assay, Karl Fischer titration, ion chromatography, headspace gas chromatography. If the method is not named, the value was not measured.

There is one limitation no routine method fixes. Intact mass confirms the mass of the molecule, not its sequence order or its stereochemistry. Leucine and isoleucine are isomers of identical mass. A scrambled sequence built from the same residues has the same mass as the correct one. A peptide carrying a D-amino acid where an L-amino acid belongs has the same mass as the correct one. Resolving those requires tandem MS sequencing or amino acid analysis, which most routine COAs do not include.

Document fraud in this space is rarely sophisticated. It is usually a matter of reuse, cropping, and typography, and it falls apart under a few minutes of attention.

The most common pattern is the recycled chromatogram. One clean trace gets reused across lots and sometimes across compounds. Because instrument noise is effectively random, two genuine chromatograms never share identical baselines, so if you overlay two documents and the baseline wiggles match, they came from one file. Retention times identical to three decimal places across different compounds are the same tell.

The second pattern is the cosmetic COA: a summary page reporting a purity figure and a mass with no instrument output at all, so nothing on it can be checked. The third is selective cropping, where the image is trimmed so the early or late region of the run is not visible. A trace that begins four minutes into a twenty minute gradient has had its solvent front region removed.

Reviewing a COA well takes about five minutes once the routine is familiar. Work top to bottom, and if any step fails, ask. A supplier working from real analytical data can produce the raw instrument file, name the laboratory and the method, and arrange a retest of the same lot. A supplier working from a template cannot, and that answer is usually more informative than the document that prompted it.