Home / Science / How peptides are made
Veriphase · Explained simply
Almost every quality problem in this market traces back to one process. Understanding it takes about six minutes and explains the rest of this section.
A peptide is a chain. Amino acids are the links. Making one means joining those links in an exact order — and getting every single one right.
That sounds simple, and the principle is. What makes it hard is that you cannot see what you are building, cannot pick it up, and cannot check your work until the very end.
For most of the history of peptide medicine, nobody made peptides at all. They took them out of animals.
Two physiologists at University College London, William Bayliss and Ernest Starling, were investigating how the pancreas knows to start working when food arrives. The prevailing assumption was nerves.
Working on an anaesthetised dog, they cut the nerves to a section of small intestine, then introduced dilute acid into it — imitating what arrives from the stomach. The pancreas secreted anyway. Whatever carried the message had travelled through the blood.
They called the substance secretin. It was the first hormone ever identified, and there was at the time no word for what it was. Starling proposed one in 1905: hormone.
Secretin is a peptide. Twenty-seven amino acids in a chain.
So the first hormone anyone discovered, and the molecule the entire concept of hormones was invented to describe, is the same kind of molecule sitting in the vials this laboratory tests.
Nineteen years later and nearly six thousand kilometres west, Frederick Banting and Charles Best spent a summer at the University of Toronto tying off the pancreatic ducts of dogs, extracting what remained, and injecting it into other dogs whose pancreases had been removed. The diabetic dogs recovered. The result was announced in November 1921.
Insulin is also a peptide — fifty-one amino acids across two linked chains.
On 11 January 1922, a fourteen-year-old boy called Leonard Thompson became the first person given insulin. He was dying of diabetes, and at that time everyone with the disease did.
It did not work. His blood sugar fell by about a quarter, but he kept producing ketones, and the extract caused a toxic reaction with abscesses at the injection site. The problem was not the molecule. The problem was everything else in the bottle.
The biochemist James Collip spent the next twelve days improving the purification. On 23 January Thompson was injected again with Collip’s cleaner extract. His blood sugar stabilised, the ketones cleared, and there were no ill effects.
The first peptide drug in history failed on its first attempt because it was not pure enough. Twelve days later, purified, it saved a life.
That happened in Toronto, about thirty kilometres from where your sample gets analysed. It is the oldest lesson in this field, and it has not changed: the molecule being right is not the same as the preparation being right.
Once insulin worked, it had to be produced at scale — and the only source was animal pancreases, collected from slaughterhouses, processed in industrial quantities to yield very small amounts of usable extract. Every diabetic in the world depended on cattle and pigs, and on purification good enough to make what was extracted safe.
That is what the whole industry was until the 1960s. Not chemistry. Butchery, followed by careful cleaning.
What changes everything is the realisation that you do not have to find a peptide at all. You can assemble it, link by link, from nothing.
Which is easier said than done. Imagine assembling a necklace of forty specific beads in a specific order, in a bucket of water, in the dark. Every time you add a bead you have to fish the necklace back out to check you have not lost it, and rinse away all the spares before adding the next one.
In 1963 a chemist called Bruce Merrifield had the idea that solved it, and it won him a Nobel Prize twenty years later. Glue one end of the chain to something you cannot lose.
Anchor the first amino acid to a tiny solid bead. Now the growing chain is stuck to something you can hold on to.
Add the next amino acid in excess, let it react, then simply wash the bead. Everything that did not attach flows away. What is left is your chain, one link longer.
Repeat. That is solid-phase peptide synthesis, and essentially every peptide you can buy is made this way.
Each new link means running the same short cycle. For a 39-residue peptide, thirty-eight times.
At the end the finished chain is cleaved off the bead, the remaining protective groups are stripped, and you have crude peptide.
The arithmetic behind most of what goes wrong is not complicated.
Suppose each coupling works 99% of the time. That sounds excellent, and on any single step it is. But the steps multiply.
| Peptide | Residues | Couplings | At 99% per step | At 99.5% | At 99.8% |
|---|---|---|---|---|---|
| Ipamorelin | 5 | 4 | 96.1% | 98.0% | 99.2% |
| BPC-157 | 15 | 14 | 86.9% | 93.2% | 97.2% |
| Semaglutide | 31 | 30 | 74.0% | 86.0% | 94.2% |
| Tirzepatide | 39 | 38 | 68.3% | 82.7% | 92.7% |
Proportion of chains complete and correct before any purification. Stepwise efficiency compounds: 0.99 raised to the number of couplings.
At 99% per step, roughly two thirds of a 39-residue chain comes out right. The other third does not.
So a batch of crude tirzepatide is not tirzepatide with a few impurities in it. It is a mixture — correct chains alongside chains missing a link, and chains that stopped growing partway.
Those near-misses have names. A deletion sequence is a chain where one coupling failed and the process carried on regardless. A truncation is one that stopped early. Both are chemically very similar to the real thing, which is exactly why separating them takes real equipment.
Length. Obvious from the table — more links, more chances to fail. A five-residue peptide is comparatively easy. A thirty-nine-residue one is not.
Sequence. Not all links are equally cooperative. Proline in particular makes the growing chain fold back on itself and hide the end you are trying to react. BPC-157 has four prolines in fifteen residues, which is why sequence fidelity is the thing worth checking there rather than outright substitution.
Modifications. The GLP-1 peptides are not just chains. Tirzepatide carries a fatty acid attached through a spacer at one specific position, added after the chain is built. That step has its own failure rate, and when it fails you get molecules with the right sequence and no attachment.
Because the crude product is a mixture, it has to be separated. That means chromatography — pushing everything through a column so the components emerge at different times, and collecting only the fraction you want.
The liquid pushed through that column almost always contains trifluoroacetic acid, usually shortened to TFA. It makes the separation work, and it is also used in the step that cuts the finished chain off the bead.
And it does not entirely leave. TFA pairs with the charged parts of the peptide and stays with it through drying, ending up in the vial as part of the powder that gets weighed.
Purification can produce a genuinely pure peptide — the right chains, separated from the wrong ones.
It also leaves salt behind. So a vial can be exactly what it claims, at excellent purity, and still contain meaningfully less peptide than the label states. That gap has its own article. This is where it comes from.
The purified peptide is in solution, and solutions are unstable and awkward to ship. So it is frozen and the water pulled off under vacuum — lyophilisation — leaving the white powder that arrives in a vial.
Two properties of that powder matter. It is hygroscopic: it pulls moisture out of the air, so some of what you weigh is water. And every stage from synthesis onward — the water used, the glassware, the air in the room — is an opportunity for bacterial endotoxin to get in, which is why that is a separate test with a separate answer.
Synthesis was not the end of it. In 1978 human insulin was produced for the first time by engineered bacteria, and by 1982 recombinant human insulin was the first genetically engineered medicine approved anywhere. Larger peptides and proteins are now usually grown rather than built.
But for peptides of the size discussed on this site — five residues to forty — chemical synthesis on a bead remains the standard method. The process described above is how the contents of your vial came into existence.
Almost everything else in this section traces back to the process above.
Why purity varies between vendors. Stepwise efficiency and purification quality are where the money goes. A factory running at 99.8% per coupling with careful chromatography produces something genuinely different from one running at 99% — and the difference is invisible in a powder.
Why identity failures happen at all. A deletion sequence is not a forgery. It is a real molecule from a real process that went slightly wrong, which is why catching it means weighing the molecule rather than trusting a label.
Why the amount is so often wrong. Salt, residual water and imperfect fill weights all sit between “we made this peptide” and “there are 10 mg of it in here.” Across the largest dataset available, that turned out to be the most common failure of all.
None of this requires anyone to be dishonest. It is what happens when a difficult manufacturing process meets a market with no mandatory quality control. The chemistry is unforgiving and nobody is obliged to check.
Verify always.
For research use only. Not for human or veterinary administration, and not a clinical or diagnostic service. Yield figures are illustrative calculations of stepwise efficiency, not measurements of any specific product.