A large part of this catalog is not a whole molecule. It is a piece of one.
Some entries announce it — HGH Fragment 176-191, Kisspeptin-10. Others conceal it completely: Ara-290, TB-500, KPV, Thymosin Alpha-1, MGF, Sermorelin, Semax, AOD-9604. All exist because somebody took a larger protein and kept part of it.
That is a research class, and the one class this catalog has no shelf for. It cuts across metabolic, growth-hormone, healing-recovery, immune, cognitive-mood and sexual-health filings, because it is defined by a design decision rather than a receptor or an endpoint — and it carries one characteristic reading error: treating "derived from X" as a claim that the fragment does what X does.
Fragments of fragments
The nesting goes deeper than the names suggest. KPV is the C-terminal tripeptide of alpha-MSH — and alpha-MSH is itself a fragment, cut out of proopiomelanocortin. Kisspeptin-10 is the C-terminal decapeptide of kisspeptin-54, which is itself a fragment of a 138-residue precursor (UniProt Q15726: kisspeptin-54 at 68-121, kisspeptin-10 at 112-121).
Thymosin Alpha-1 corresponds to residues 2-29 of prothymosin alpha. Sermorelin is GRF(1-29)-amide, the N-terminal 29 residues of a 44-residue hormone which is itself excised from a 108-residue precursor.
So "fragment of X" is a statement about origin, not about equivalence. Across this class, three different relationships to the parent are possible, and which one applies is an empirical question answered separately for every compound.
Group one: the fragment is the message
Sometimes the retained piece contains the binding determinants, and the parent's receptor pharmacology largely transfers.
GRF(1-29)-amide is the textbook case. Peticlerc et al. (J Anim Sci 1987;65:996-1005) ran a dose-response comparison of hGRF(1-44)NH2 against hGRF(1-29)NH2 in heifers and pigs in a double Latin square design, and found the two equally stimulated growth hormone release across the tested range in both species, concluding the two are equipotent. Bongers et al. (Pept Res 1992;5(4):183-9) refer to GRF(1-29)-NH2 straightforwardly as "the equipotent analog" and report full potency in an in vitro rat pituitary bioassay alongside full receptor-binding affinity. The C-terminal 15 residues of the hormone are not carrying the message.
Kisspeptin-10 is similar with an instructive wrinkle. Mikkelsen et al. (Regul Pept 2009;152:95-100) compared peripherally administered kisspeptins of different lengths in adult male mice: mouse kisspeptin-10 and -52, and human kisspeptin-10 and -54, all raised serum testosterone, and the short and long forms showed similar potency and efficacy. What differed was the time course — the longer forms had slower onset, and the long human form a more prolonged effect.
This site has repeatedly flagged kisspeptin-54 and -10 as non-interchangeable. That flag stands, but this refines it: the published difference is kinetic, not a difference in potency at the receptor — exactly the distinction the pharmacokinetics primer and receptor pharmacology primer exist to separate. A fragment can match its parent's pharmacology and still occupy a different timescale.
Group two: the cut is the point
For a second group, failing to reproduce the parent is the entire design intent. Reading the parent's file onto these compounds inverts the rationale.
- AOD-9604 and HGH Fragment 176-191 isolate the C-terminal lipolytic region of growth hormone specifically without the growth and IGF-1 activity of the intact hormone.
- Ara-290 is derived from the helix-B surface of erythropoietin — the aqueous-facing side of the helix, pointing away from the receptor when EPO is docked. The stated goal was tissue protection separated from erythropoiesis.
- Ac-LKKTETQ corresponds to the actin-binding motif of thymosin beta-4, not to the whole protein.
- KPV discards the entire His-Phe-Arg-Trp melanocortin pharmacophore. It retains anti-inflammatory activity, has no pigmentary action, and its mechanism appears not to run through melanocortin receptors at all.
- Semax is ACTH(4-7) with an added Pro-Gly-Pro tail; the corticotropic activity of ACTH is removed by construction.
For this group the honest statement is not that the parent's evidence may not transfer, but that the fragment was built so that it would not.
Group three: where the parent link is the only established thing
A third group has a documented ancestry and very little else.
MGF is the clearest case. The E-domain is a pro-peptide removed during IGF-1 processing, containing none of the mature IGF-1 sequence that binds IGF-1R. As the MGF deep-dive covered, the proliferative effect failed to replicate in an industry-run study, and no analogous free E-peptide has been isolated as a separate molecular species. The parentage is real; it carries no pharmacological commitment.
Thymosin Alpha-1 shares a surname with thymosin beta-4 purely because both came out of the same crude thymic preparation and the Greek letters were assigned by isoelectric point. As the comparison with Thymalin set out, that name records how something was found, not what it does.
What the 2026 literature says about fragments in general
Two independent 2026 papers make the general argument better than any single-compound file can.
Ricard-Blum and Fradette (FEBS J 2026;293(13):4064-4084, PMID 41665202) reviewed the landscape of collagen bioactive fragments and state the class thesis directly: fragments released on extracellular-matrix remodeling are bioactive with functions that can differ from those of their parent proteins, most often in angiogenesis, inflammation, fibrosis and cancer. Same logic underlies the matrikine framing of GHK-Cu, with that post's caveat: a matrikine is a product of remodeling, and reading it as an instruction is an inference.
The sharper result is Gaglione et al. (JACS Au 2025 Dec 10;6(1):124-143, PMID 41614195), who characterized three antimicrobial encrypted peptides taken from human matrix metallopeptidase-19 (residues 1-19, 1-33 and 247-279). They report broad-spectrum activity against Gram-positive and Gram-negative bacteria including multidrug-resistant strains, with membrane depolarization and permeabilization as the primary mechanism.
The parent protein is a protease. Nothing about its enzymology predicts or explains fragments that kill bacteria by membrane physics. Cutting a protein can reveal an activity the intact protein does not have.
Scale comes from Costa et al. (FEBS J 2024;291(3):547-565, PMID 37945538), who found cationic amphiphilic segments in around 2% of all human proteins — and reported that activity of the synthetic sequences was irrespective of whether those segments are membrane-associated in the cellular setting. A synthesized fragment can be active without that activity corresponding to anything the parent does in a cell.
All of this is rung-one biochemistry on compounds not in this catalog — the wider proteome-mining field is reviewed by Crysler and de la Fuente-Nunez (Cell Host Microbe 2026;34(7):1273-1284, substantial commercial interests disclosed). It establishes a direction of reasoning, not a result about any product here.
The number in the name needs a frame
Fragment names use at least three incompatible conventions, and the numeric ones are unreadable without knowing the coordinate frame.
- Growth hormone: UniProt P01241 is a 217-residue precursor whose mature chain runs 27-217, giving the familiar 191 residues. "176-191" is mature numbering. The same residues are 202-217 of the precursor.
- GHRH: UniProt P01286 is a 108-residue precursor with the mature 44-residue hormone at 32-75. "GRF(1-29)" is mature numbering; in precursor coordinates, 32-60.
- EPO: UniProt P01588 is a 193-residue precursor with the mature chain at 28-193. Ara-290's "helix B, residues 58-82" is mature numbering; UniProt's own helix annotation for the same structural element reads 83-109. Same helix, two sets of numbers — and Ara-290's name contains no number at all.
- Kisspeptin-10: here the numeral is a length, not a coordinate. Kisspeptin-14, -13 and -10 are nested and all end at residue 121.
This extends, rather than repeats, the nomenclature primer and the sequence primer: the failure mode is not just confusing a coordinate with a length, it is comparing two numbers drawn from different frames.
A fragment has two new termini
This is the analytical consequence that matters most, and it follows from one observation: a fragment is not a subset of its parent. It is a new molecule carrying chemical features the parent did not have at those positions.
Kisspeptin-10 shows both halves at once. Its C-terminal phenylalanine is amidated — UniProt annotates residue 121 as "Phenylalanine amide," and that amide is shared by kisspeptin-54, -14, -13 and -10 because it is the real hormone terminus. That one is inherited. Its N-terminal tyrosine, by contrast, was an interior residue of kisspeptin-54 and only became a free amine because of where the cut fell. That one is manufactured.
Across the class, the manufactured terminus is where the chemistry lives:
- KPV's C-terminal valine is amidated in alpha-MSH, but the CAS number on catalog records resolves to the free acid — about 1 Da and one unit of net charge apart, a question created entirely by where the sequence ends.
- Ara-290 carries an N-terminal pyroglutamate, roughly 17 Da from free glutamine and a group the parent does not have at that position.
- Ac-LKKTETQ carries an N-acetyl cap, +42 Da, added because the cut exposed an amine.
So the theoretical mass on a fragment's paperwork must be the mass of the fully annotated fragment, and the termini are precisely where annotations get dropped. A bare-letters calculation produces a mismatch that looks like a product problem and is actually a paperwork problem — the same failure mode described in reading a chromatogram and mass spectrum.
And the new terminus is a peptidase entry point, which is why so many fragments on this shelf are capped at all. Worth noting that fragments are themselves substrates: UniProt annotates a cleavage site at residue 118 — inside kisspeptin-10 — attributed to MMP2 and MMP9. More on the enzymology in the peptidase and clearance explainer, and on the charge consequences in the charge and solubility primer.
FAQ
Is a fragment weaker than its parent? Potency is not a function of length. GRF(1-29)-amide is reported equipotent with the 44-residue hormone; AOD-9604 was built to keep one activity and drop another; MGF's basic activity is disputed. Three fragments, three answers.
Can the parent's trials be cited for the fragment? This is the characteristic error of the class. The answer depends on which group the compound falls in, and for two of the three groups it is no. Thymosin beta-4's human trial record is on full-length thymosin beta-4, not the heptapeptide. Erythropoietin's approval file is not Ara-290's file. See the evidence hierarchy.
Why treat this as a class when the catalog has no such shelf? Because the shelves sort by intended endpoint, while this property cuts across all of them and predicts a specific reading error. Sorting by where the cut was made tells you which questions are already answered by the parent's literature and which are not answered at all. More at /quality/ and the /library/.
This article is educational and for the laboratory research community. Trulogic Labs products are sold for laboratory and research use only and are not for human consumption.