How a score is calculated
Every score is deterministic: the same certificate and the same reference data produce the same number, every time. There is no model judgement in the score and no human adjustment. Each verdict page carries the audit log that produced it, so a dispute is a factual question rather than an appeal to opinion.
The bands
What is weighted
Two dimensions carry the score. The weights are relative and are renormalized over whichever dimensions the certificate actually contains, so a COA that omits a contamination panel is not silently punished as if it had failed one:
- Purity — weight 0.5. Present on every certificate. Reads the reported HPLC purity, and the chromatogram shape where one is legible.
- Contamination — weight 0.3. Heavy metals, residual solvents, endotoxins. Scored only when a panel is actually on the certificate.
What is never scored
Lab origin carries a weight of zero. The testing lab named on a certificate is recorded and shown, but it does not move the number, because Peptora does not independently verify laboratory accreditation. Scoring a lab we have not audited would be inventing authority we do not have.
Nothing is scored on commercial terms. No vendor can pay to be ranked, delisted, or re-scored, and no placement in the Trust Index is for sale.
The gates
Some findings are not a slice of the average — they cap the result outright, because a strong number in one place cannot vouch for a failure in another.
- Molecular-weight mismatch. If the mass printed on the certificate disagrees with the mass computed from the printed sequence, the score is capped regardless of purity. A very pure sample of the wrong molecule is not a good result.
- Unconfirmed identity caps at 84. When a compound matches a catalogue entry by name but the reference mass is a best-known published value rather than an independently verified one, presence is confirmed but identity is not. 84 sits one point below the Verified threshold on purpose: the certificate is not suspect, it simply has not proven what “verified” would claim.
- Blends cap at Provisional. On a multi-peptide certificate only the presence of each named peptide is checked; the proportions are not independently confirmed, so a blend can never reach Verified.
- Document reuse across sellers. The same certificate appearing under two unrelated sellers is a failure, not a warning.
Not measured is not the same as clean
A certificate that never tested for heavy metals does not get credit for passing that test. Peptora shows what was measured and names what was not, rather than letting silence read as a pass — which is the single most common way a thin certificate looks like a good one.
Where the certificates come from
Some records begin with a user upload. Others are collected from vendors’ own public COA pages so the index is useful on day one; the limits that crawler runs under, and how a vendor stops it, are documented at the collector.
Contesting a verdict
If a published verdict concerns your product and you believe it is wrong, contest it. Because every score is reproducible from its audit log, a dispute resolves on the facts. When a finding is wrong we correct it and say so on the record.
Peptora reports on documents and data for research-use-only materials. It makes no human-use, dosing, or therapeutic claims.