How the WhichPeps evidence system is structured.
The site is the interface. The structured evidence records, provenance and update process are the reusable asset underneath it.
The public website is generated from structured identity, evidence, safety, regulatory and editorial fields. Learn pages explain those fields; compound/study pages expose them; future professional outputs should use the same schema rather than creating a second scientific truth.
Current data relationships
| Field group | What is tracked | Why it matters |
|---|---|---|
| Compound identity | Canonical ID/name, display name, aliases, brands/codes where relevant, entity type, related entities. | Prevents duplicate or conflated evidence records. |
| Evidence record | Study/source ID, evidence type, species/population, design, sample size, formulation, route, dose studied, duration, endpoints, results, limitations. | Keeps a headline claim attached to the context that produced it. |
| Safety | Adverse-event terms, source, population/context, seriousness where reported, uncertainty. | Avoids implying absence of evidence is evidence of safety. |
| Regulatory | Current launch record: canonical ID/name, a dated regulatory status summary, last-reviewed date, canonical profile URL and aliases. Product, jurisdiction, indication and regulator detail is retained in linked sources and profile wording where relevant, but is not yet normalized into separate relationship fields. | Makes the current data boundary explicit and avoids implying a richer normalized export than the launch dataset contains. |
| Editorial | Evidence strength, reviewed date, updated date, corrections and change reason. | Makes freshness and editorial changes auditable. |
Entity resolution
One canonical ID represents one scientific entity. A brand name, development code, nickname or common spelling variant can be searchable without creating a duplicate evidence record. Formulation differences are preserved where scientifically or regulatorily meaningful.
Source selection and evidence classification
WhichPeps prioritises primary peer-reviewed studies, trial registries and official regulator sources. Human, animal, laboratory, review and regulatory records remain distinct. Mechanism evidence is not promoted to clinical-benefit evidence simply because it is biologically plausible.
Review dates and updates
Evidence reviewed records when the scientific record behind a profile was reviewed. Page updated records when the webpage/build changed. These dates are intentionally separate.
Corrections and change history
Material errors or new evidence can change a structured field or public summary. The correction pathway and public change log provide a place to record those changes without silently replacing the historical context.
Sample structured record
{
"id": "RETATRUTIDE",
"name": "Retatrutide",
"status": "Not FDA-approved as of this review; regulatory context is summarized in prose and linked from the profile.",
"lastReviewed": "2026-09-20",
"url": "/compound/retatrutide/",
"aliases": "Reta; LY3437943; GLP-3; Triple G"
}This preview mirrors the current launch-facing regulatory record shape. Product, jurisdiction, indication and regulator relationships are not presented here as normalized fields unless the underlying dataset actually stores them separately. No public API currently exists.
Data-governance rules
- One canonical ID per entity; aliases do not create duplicate scientific records.
- Unknown fields remain unknown/not established rather than being filled for appearance.
- Evidence-reviewed date and page-update date are separate.
- Automation may find, format, compare and draft, but scientific meaning is not auto-published without human approval.
- Future exports must come from the same schema used to build public pages.