What each verdict actually means
Four tiers, shared across the network, and the rules each agent uses to move a result from one to the next.
The four tiers
Every agent's verdict lands in one of these, and the words mean the same thing on every subdomain. That is what lets one result page render any agent's answer.
| Tier | Label | What it means |
|---|---|---|
| 1 | Strong | Direct evidence for this exact subject, from a named source. |
| 2 | Moderate | Real evidence, but indirect: a close neighbour, an analog, or a partial match. |
| 3 | Weak | Evidence of a problem, or evidence too thin to rely on. |
| 4 | Insufficient | No evidence either way. Not a failure, and never styled as one. |
Insufficient is deliberately muted rather than alarming. Absence of evidence is not evidence of harm, and an agent that styled "no data" as a failure would be making a finding it did not make.
Per agent
No grading method published yet.
LiveThe top line is the worst finding, and it names which one. Five dimensions are scored independently and never averaged. The headline tier is the worst of them, and the sentence beside it names every dimension at that tier. 'Weak' alone is a grade over the result; 'Weak: 2 dimensions at this level, aggregation propensity and proteolytic stability' is a pointer at two rows further down the page.
Insufficient is excluded from the worst-of. Tier 4 is not worse than tier 3. It is a different axis. Treating 'we could not check' as 'we checked and it is bad' would report a missing database as a finding about your peptide. Every tier-4 dimension becomes an entry in the limits block instead, carrying its own reason.
The same word can mean opposite things. Self-assembly 'low' is a genuinely weak motif match, which is a negative finding. Aggregation 'low' means only that no hydrophobicity-driven signal was detected, by a scale that is blind to polar-zipper aggregation, so it maps to insufficient rather than strong. Reading the tier off the word would colour a blind spot green.
Two dimensions can never reach tier 1. MEROPS records cleavage events only, so it can show a peptide is vulnerable and can never show it resists. CycPeptMPDB matches close analogs, never this molecule. Both are capped below strong by design, and the grading method says so rather than letting a reader wonder why they never see a green row there.
No grading method published yet.
BetaNo grading method published yet.
BuildingNo grading method published yet.
BuildingNo grading method published yet.
Planned