Methodology
How this site is made
Every profile and goal brief here is assembled from cited research, graded for the strength of that research, checked claim by claim against the sources, and read by a person before it is published. This page explains each of those steps, and what they do not guarantee.
How a profile is made
For each compound we gather sources from PubMed and the open web — peer-reviewed papers, regulatory labels, reviews, and the vendor pages and community reports where most people first hear about a peptide. Each source is graded on the scale below before anything is written from it.
From every source we extract its findings as individual, attributed statements, each labelled with the kind of evidence behind it: a human trial, a human observational study, an animal study, cell-culture work, expert opinion, or anecdote. The profile's prose is then written from those findings — not the other way round — which is why every sentence on a profile can point to the source it came from. The numbered marks in the text are those pointers.
Nothing publishes itself. A draft passes the checks described below and is then read, corrected where needed, and published by a person.
How we grade evidence
Four tiers, shown as a four-segment meter throughout the site. More filled segments means stronger evidence — and Tier 4 still fills one, because a plausible mechanism is not nothing.
- Tier 1 · Human trials
- Human randomized controlled trials, or strong observational data in people.
- Tier 2 · Preclinical
- Promising animal or cell-culture data. The mechanism is established; whether it holds in humans is not. Most research peptides sit here.
- Tier 3 · Reported use
- Community-reported effects with no controlled data — acknowledged openly, and labelled as what they are.
- Tier 4 · Theoretical
- The mechanism exists on paper. There is no meaningful experimental data yet.
The tier grades the evidence about this compound. Evidence does not transfer between compounds: a strong trial of a related molecule, a different route of administration, or a parent drug does not raise the tier of the compound on the page. A blend does not inherit the tier of its ingredients — the research usually cited for a blend is research on one of them, so a blend page shows each ingredient's own meter instead.
Two things the tier does not tell you: how much evidence there is, and whether any of it applies to your situation. The profile's findings section counts what is behind each claim; the tier is the strength of the best of it.
What the checks look for
Before a person reads a draft, it is checked automatically, and a draft that fails a check is held.
- Every claim against its source. Each statement is compared with the text of the source it cites. A claim the source does not support, or contradicts, is not published as written.
- Reporting, not advising. The site reports what studies, labels and reviews say, attributed to them. Text that speaks in its own voice about treating a condition, or tells a reader what to do, is held.
- Doses name their source.Where a regulatory label exists, its dosing is quoted as the label's. Where none exists, the profile says so first, and every figure that follows names where it was reported and how strong that report is.
- Retractions. Cited papers are checked against the retraction registry. A retracted paper is not deleted from a profile — it stays in the bibliography, labelled, and so does every finding that rests on it, so a reader can see what changed.
These checks are sentence-level. They can tell whether a sentence is supported by its source; they cannot tell whether a source is good science, and they cannot judge your circumstances. That is what the tier, the attribution, and the disclaimer on every page are for.
Points of contention
Where the evidence is unsettled or says less than the popular claim, the profile says so in its own section rather than smoothing it over.
Contested
Limited evidence
Relationships and the stack analyzer
Compounds are linked by their mechanisms — receptor targets, pathways, effects, organ systems — from a controlled vocabulary. Each link is proposed from the compound's own profile, checked against it, and approved by a person; a link the profile does not establish is left out.
The stack analyzer reads those links and the research on combining compounds, and reports what it finds: where two compounds are redundant or better kept apart, where the research describes them as complementary, and where there is no documented interaction at all. It reports; it does not recommend. A caution is kept even when the evidence for it is thin, because erring toward caution protects a reader; a reassurance is shown only when the research supports it. Overlap between two compounds is a fact about their mechanism tags, not a verdict that one is unnecessary, and “no documented interaction” is never a safety clearance.
What this site does not do
Peptide Atlas does not sell anything, does not rank or recommend vendors, carries no advertising, and gives no medical advice. It is a reference: the research, graded and attributed, organised by what people are trying to achieve. What you do with it is between you and a licensed clinician.
Found an error, or a source that says something other than what a page reports? See About for how to reach us.