What a Peptide Is, and What the Word Hides

The definition, and where it stops being useful

A peptide is a chain of amino acids linked by peptide bonds. That is the whole definition.

What it does not include is a length. Convention puts the boundary between "peptide" and "protein" somewhere around 50 amino acids, and the convention is soft enough that insulin — 51 residues — is routinely called both.

So the word covers a range of molecules with almost nothing in common except a bond type. Here is what that range actually looks like, and where the confusion does damage.

The range, from labels

Every figure below is quoted from the DESCRIPTION section of an FDA label, retrieved in August 2026.

Liraglutide — molecular weight 3,751.2 daltons. Its peptide precursor is produced by recombinant DNA expression in *Saccharomyces cerevisiae*, engineered to be 97% homologous to native human GLP-1 by substituting arginine for lysine at position 34¹.

Semaglutide — molecular formula C187H291N45O59, molecular weight 4,113.58 g/mol. The peptide backbone is produced by yeast fermentation².

Tirzepatide — molecular weight 4,813.53 Da, empirical formula C225H348N48O68, based on the GIP sequence³.

Dulaglutide — a fusion protein of two identical disulfide-linked chains, each an N-terminal GLP-1 analog sequence covalently linked to the Fc portion of a modified human IgG4 heavy chain, produced in Chinese hamster ovary cells. Overall molecular weight approximately 63 kilodaltons⁴.

Dulaglutide is roughly seventeen times the mass of semaglutide, and it is discussed inside the same drug class. Whether it is a "peptide" is a question the word cannot answer.

These molecules contain amino acids that do not exist in you

This is where the intuition "a peptide is just a small protein" breaks completely.

The tirzepatide label states that the molecule contains aminoisobutyric acid at positions 2 and 13, a C-terminal amide, and a lysine at position 20 attached to 1,20-eicosanedioic acid via a linker³.

Aminoisobutyric acid is not one of the twenty amino acids your genetic code specifies. Eicosanedioic acid is a twenty-carbon fatty diacid, not an amino acid at all.

A modern peptide drug is a designed molecule with a peptide backbone, non-natural residues, and chemical decorations that have nothing to do with biology. It is not a hormone with a slightly different spelling.

Confusion one: a fragment is not its parent

Thymosin β4 is a 43-amino-acid protein. TB-500 is a fragment of it — the registry record for the single trial registered under that name in August 2026 describes it as the "Thymosin Beta 4 17-23 Fragment"⁵.

Residues 17 through 23 is seven amino acids. It is not the same molecule, and evidence generated on one does not transfer to the other.

This matters because the human trial literature is almost entirely on the parent. We take the counts apart in healing peptides and the gap between anecdote and trial, and the compound record sits on the TB-500 page.

Confusion two: some "peptides" are not peptides

Two of the compounds most often discussed inside peptide conversations are small molecules.

MK-677, also called ibutamoren, is described in the paper that first reported its anabolic effect as an orally active *nonpeptide* mimic of a growth-hormone-releasing peptide⁶. It acts on the same axis as several peptides. It is not one. See growth-hormone secretagogues.

Orforglipron is a small-molecule, nonpeptide oral GLP-1 receptor agonist⁷. It hits the same receptor as semaglutide from a different class of molecule entirely — which is precisely why it can be swallowed without an absorption enhancer. See oral vs injectable peptides.

The practical consequence: "peptide" in a marketing context is often a claim about a category feeling, not about chemistry. Checking the actual molecule takes one look at a label's description section.

Confusion three: naming conventions imply relationships that do not exist

BPC-157 is described by a 2026 review as a synthetic pentadecapeptide — fifteen amino acids — derived from a gastric protein fragment⁸. Tesamorelin's label describes it as a synthetic human growth hormone-releasing factor analog comprising the 44-amino-acid sequence of human GRF plus a hexenoyl moiety, a C6 chain with a double bond at position 3, attached to the N-terminal tyrosine, with a molecular weight of 5,135.9 Da⁹.

Those two things share a bond type and nothing else. The "-157" is a laboratory designation and indicates no family, no mechanism, and no level of evidence.

Every compound in this category has to be evaluated on its own record. The suffix tells you nothing.

Why the category exists at all

The reason the word survives despite covering so much ground is that peptides occupy a genuine middle ground in drug design: large enough to bind protein surfaces that small molecules cannot, small enough to synthesise chemically rather than growing in cells.

That middle ground has produced a real body of medicine. A 2021 review in *Nature Reviews Drug Discovery* counts more than 80 peptide drugs reaching the market since the introduction of insulin almost a century ago, across diabetes, cancer, osteoporosis, multiple sclerosis, HIV infection and chronic pain¹⁰.

Eighty-plus approved drugs is a serious record. It is also, notably, a record built almost entirely by injection — which is what the oral vs injectable page is about.

The three questions worth asking about any "peptide"

How long is it, and is it the whole molecule or a fragment?

Is it a peptide at all, or a small molecule discussed alongside them?

And what has been measured in humans, as opposed to what has been measured in a dish? Our research index takes that question compound by compound.