How we built the literature archive.

How papers were selected

We start every category from primary sources, not from a list of peptides we think the site should cover. For each peptide that sits in the rest of the catalog, we look for the strongest human-trial evidence available; where no human-trial evidence exists, we look for the foundational mechanism papers and say so in the summary.

For each candidate paper we resolve to a stable URL in this order of preference: the publisher DOI, then PubMed, then PubMed Central, then a preprint server. Every URL is opened and confirmed to resolve before the entry is published. A paper that we can’t verify isn’t in the archive, even if we’ve read it on paper.

Inclusion rules:

• Indexed in a peer-reviewed venue, or a Cochrane review.
• Reports primary data, a major systematic review, or a foundational mechanism result that downstream papers depend on.
• First author and study group are independently identifiable (not anonymous or sponsor-authored marketing material).

Exclusion rules:

• Vendor blog posts and compounding-pharmacy marketing pages.
• “Top 10”-style listicles, even from otherwise reputable outlets.
• Predatory journals, judged per discipline.
• Animal-only papers presented as if they were clinical evidence. We tag them preclinical and call out the species in the summary instead.

What the metadata fields mean

Every card on the archive carries a small set of fields. They aren’t decorative — read them before the summary.

studyType

Six values, in roughly descending evidential weight for a human therapeutic question:

• clinical-trial— a prospective study in humans with a defined protocol. The strongest tier.
• meta-analysis— pooled results across multiple human trials.
• observational— human data without randomization (cohort studies, registries).
• review— a synthesis paper, not new data. Useful as a map; not itself a finding.
• mechanism— work on how a peptide acts at the cellular or biochemical level. Often preclinical.
• preclinical — animals only. No humans involved. This distinction matters more than any other tag on the site. A striking result in rats is not a clinical result. A striking result in cell culture is not even that.

Several peptides in this archive — BPC-157, dihexa, pentadeca-arginate — have no peer-reviewed human clinical trials at the time of writing. If you came here expecting human RCT evidence and found preclinical, that’s the field telling you something.

prominence

A three-tier editorial rating that drives how prominently an entry sits on category pages:

• essential— the paper to read if you read one in this category. Usually one or two per category.
• notable— landmark trials, foundational mechanism papers, the second-tier must-knows.
• standard— solid evidence, worth knowing, no badge.

We keep essential lean on purpose. If a fifth of the archive earned the top badge, the badge would stop meaning anything.

productCategorySlugs and peptideSlugs

An entry can belong to more than one product category (e.g. a paper on GHK-Cu lives in both recovery and longevity) and can tag more than one peptide. These slugs are how a paper surfaces on the relevant category page and in the filtered /literature?category= views.

Why the count differs across peptides

We don’t pad. The number of entries in each category reflects how much published literature exists, weighted by how much of it is informative.

• Semaglutide and tirzepatide dominate the GLP-1 category because they dominate the actual research output. Most landmark trials of the past five years involve one of them.
• Epitalon, BPC-157, GHK-Cuhave fewer entries despite real interest, because the indexed literature on each is dominated by a single research group — see the next section.
• Pentadeca-arginate (PDA)has essentially no indexed peer-reviewed clinical literature as of writing. We’d rather acknowledge that than fabricate substance.

Absence isn’t a curation choice. It’s the state of the field.

Author and group concentration

Three researchers appear repeatedly in this archive:

• Vladimir Khavinson (St. Petersburg) on epitalon and short-peptide bioregulators.
• Predrag Sikiric (Zagreb) on BPC-157 mechanism and tissue protection.
• Loren Pickart on GHK-Cu, going back to the 1970s.

In each case, the originating group isessentially the field. Independent replication outside the founding laboratory is limited or, in places, absent. We include more than one paper from each because excluding them would leave the category empty — but treat the cluster as one well-documented signal, not three independent confirmations.

This matters more for some peptides than others. Pickart’s GHK-Cu mechanism has been corroborated by independent groups in dermatology and gene-expression studies. Khavinson’s telomerase claims for epitalon have not been replicated outside his institute in the way the foundational paper would predict. We don’t smooth over the difference; we name it in each summary.

Diversity rules in the curation

To prevent any single group from skewing the archive:

• No more than three papers from the same first author, except in the Khavinson / Sikiric / Pickart cases above.
• Reviews are weighted toward high-impact venues — Cochrane, NEJM, The Lancet, Nature, JAMA. Lower-tier reviews appear only when they’re genuinely the best wide-angle source for a topic.
• Preprints are excluded unless the preprint is the only primary source for a landmark result already affecting the field. When included, preprint status is flagged in the summary.

What’s deliberately excluded

• Vendor blog posts, compounding-pharmacy marketing pages, and “studies” published by supplement retailers.
• Listicle articles (“Top 10 peptides for [outcome]”).
• Predatory journals. Judgment is per discipline — some open-access publishers are reputable in some fields and not others.
• Animal-only papers framed as clinical evidence. We tag them preclinical and name the species.

Known biases

Transparency means naming the limitations we know about, not just the ones a critic would find:

1. English-language bias.Most published Semax, Selank, and Cerebrolysin clinical work appeared originally in Russian. We index the English-indexed primaries plus the major Cochrane reviews. This almost certainly under-weights the actual published evidence base for those peptides — and at the same time, the Russian-language clinical literature has its own well-known methodological and replication concerns. Both things are true.
2. Recency overweight on GLP-1s. The category is the most active research area in clinical medicine right now. We include the landmark trials but deliberately cap entries to avoid drowning the rest of the archive.
3. Mechanism over phenomenology.Where a peptide has both interesting mechanism papers and underwhelming clinical results, we usually include both — but our editorial voice tends to be more skeptical of phenomenological claims unsupported by mechanism. Some readers will think this is the right calibration. Others will think it under-weights real-world clinical experience.
4. No replication-tracking field yet.The schema doesn’t currently carry a flag for whether a finding has been independently replicated. We mention it in summaries where it matters; a future version of the schema will surface this more systematically.

Tell us what we got wrong

This archive is a starting point, not a final document. If you spot:

• A paper that should be here and isn’t,
• A paper that shouldn’t be here and is,
• A DOI or URL that doesn’t resolve,
• A summary that misrepresents what a paper actually found,

email hello@savepeptides.com. We’d rather hear it from a reader than have the page lie quietly to the next one.