Methodology

PeptiDatabase collects public data on peptides from official sources, stores the exact response each figure came from, and checks every source again each day. Nothing on the site is written by hand from memory: each number, status and date is read from a source and shown with a link to it.

Sources

Source	What we take	How a record is matched to a peptide	Checked
ClinicalTrials.gov (US National Library of Medicine)	Registered studies: status, phase, conditions, sponsor, start date, whether results are posted	The peptide’s names appear among the study’s interventions	Daily
PubMed (US National Library of Medicine)	Number of publications by study type; the 200 most recent records with publication types and MeSH headings	The peptide’s names appear in the title or abstract	Daily
Drugs@FDA (openFDA)	FDA applications, products, marketing status, first approval date	Generic, brand or active-ingredient name	Daily
FDA drug labels (openFDA)	Boxed warning, warnings and adverse reactions sections. Dosing sections are not stored	Generic, brand or substance name	Daily
FAERS, FDA Adverse Event Reporting System (openFDA)	Number of reports, serious reports, reports with a death outcome, most reported reaction terms	Harmonized generic name or the product name written in the report	Daily (FDA updates the data quarterly)
FDA.gov: bulk drug substances that may present significant safety risks in compounding	Category 2 entries, withdrawn nominations and FDA’s stated risk	The name FDA lists, names in parentheses and “also known as” names	Daily

The exact search used for each peptide is visible on its page: every count links to the same search on the source.

How the data stays honest

• No source, no record. Every figure is stored together with the upstream response it was read from and the time it was last confirmed.
• Absence is recorded too. “No application in Drugs@FDA” or “not on FDA’s list” are dated results of a completed search, not assumptions.
• Failures change nothing. If a source is unreachable or a page’s structure changes, that update is discarded as a whole; the site keeps showing the last verified data with its date.
• Every change is logged. Each daily check is compared with the stored version. New, changed and removed facts appear in the changelog.

Reading the evidence

Each PubMed record is classified only from the publication types and MeSH headings assigned by the National Library of Medicine:

• Systematic review / meta-analysis: a structured synthesis of other studies.
• Randomized controlled trial (RCT): participants randomly assigned to the peptide or a control. The strongest single-study design for effects in humans.
• Clinical trial: a study in humans that is not tagged as randomized.
• Case report: a description of one or a few patients.
• Review: a narrative overview, not new data.
• Human study / animal study: other research tagged with the Humans or Animals heading.
• Unclassified: not yet indexed by PubMed, or laboratory work without a species heading.
• Retracted: the publication was withdrawn by its journal. A retracted study is never counted as evidence.

Categories, types and blends

• Research areas group substances by the field they are studied in, for navigation only. A category is our editorial classification, not a claim that a substance works for that purpose.
• Type separates peptides, larger proteins, vendor blends, and small molecules that are sold alongside peptides but are not peptides.
• Blends are vendor mixes of catalogued substances. For a blend, every source is searched for records that mention all of its components together; each component keeps its own page.
• Research topics are the most frequent MeSH subject headings in the substance’s latest PubMed records, shown as indexed by the National Library of Medicine.

News

The news page lists notices from medicines regulators and headlines from health, pharma and biotech journalism that mention a catalogued substance. Each feed is read once a day with one request, after checking the site’s robots.txt, and only feeds or announcement pages the publisher offers publicly are used.

Source	Type	Language	What we show	Basis
FDA: press announcements, MedWatch safety alerts, what’s new for drugs, recalls, health fraud	Official	English	Title, date, link and FDA’s own summary; company recall announcements that FDA reposts get title, date and link only	US federal works are in the public domain (FDA website policies); company texts are not federal works
EMA: news and press releases	Official	English	Title, date, link and EMA’s own summary, credited to EMA	Reproduction authorised with acknowledgement of the source (EMA legal notice)
AEMPS: news and information notes	Official	Spanish	Title, date, link and AEMPS’s own summary, credited to AEMPS	Reproduction authorised citing the origin (AEMPS aviso legal)
TİTCK: announcements	Official	Turkish	Title, date and link	No reuse statement on the site, so no text beyond the title
STAT, BioPharma Dive, KFF Health News, The Conversation	Media	English	Headline, outlet, date and link	Each publisher’s terms, linked on the news page
Redacción Médica, ConSalud, The Conversation	Media	Spanish	Headline, outlet, date and link	Each publisher’s terms, linked on the news page
TRT Haber	Media	Turkish	Headline, outlet, date and link	Each publisher’s terms, linked on the news page

• Media items are links, not content. We never store or show article text, feed descriptions or images, write no summaries, and do not translate headlines: each is shown in its original language and marked as such.
• Official summaries appear only where the agency’s terms permit reuse, labelled as the agency’s summary and next to its name.
• How an item is linked to a substance. The headline (and, only for matching, the feed’s short description) must name the substance by one of the names used for the other sources, or by its Spanish or Turkish spelling (e.g. semaglutida, semaglutid). Names match as whole words; a hyphen or a space makes no difference (“BPC-157”, “BPC 157”); acronym-style names match only in capitals. When names overlap, the longer one wins: “GHK-Cu” is not counted as GHK.
• Ambiguous names are never matched. Names that are ordinary words or marketing names (GLOW, KLOW, Wolverine) are excluded, and blends are linked only when their own name appears, never because their components are mentioned. Phrases such as “glucagon-like peptide” or “angiotensin II receptor blockers” do not count as glucagon or angiotensin II.
• News about peptides in general (for example “peptide”, “GLP-1”, “péptido”, “peptit”) is kept even when it names no catalogued substance: for official notices the keyword may be anywhere in the notice, for media it must be in the headline.
• Limits. Automatic matching can link a story that mentions a substance only in passing, and misses stories that use a name we do not list. Feeds carry only recent items, so anything published and dropped while a source was unreachable can be missed. Dates are the publisher’s; when a source gives none, the date we first saw the item is shown and marked. Inclusion is not endorsement or verification of what an article says, and nothing in the news feed is medical advice.

Limits

• Regulators: regulatory data currently covers the US FDA only.
• Name matching: a record filed under a name we do not search for will be missed, and a common abbreviation may match unrelated work. Synonyms are checked before they are added, and the search behind each number is shown.
• Mentions are not tests: a publication or trial that mentions a peptide may only measure it or cite it.
• Adverse event reports: FAERS reports are voluntary, can be duplicated or incomplete, and do not show that a product caused the event. Counts rise with how widely a product is used and how much attention it receives.
• Labels: openFDA also lists over-the-counter product labels that FDA has not reviewed. A label is not an approval.
• Language: trial titles and FDA statements are shown in English as published. Adverse reaction terms are translated into Turkish and Spanish with a fixed, machine-assisted glossary; the original MedDRA term is always shown under the translation.

What we never publish

Dosing, administration protocols, preparation instructions, vendors or prices.

Language selection and cookies

On your first visit to the home page, the language follows your browser’s language setting. A link to a page in another language is never redirected; you are only offered your own language. When you pick a language from the menu, that choice is stored in a single cookie (pdb_lang, 1 year). We use no advertising, analytics or tracking cookies.