# CAIN-A75-012 Research Evidence Review

**Verdict:** VERIFIED

> Recomputes a binder-evidence claim: binding hashes, replicates and controls, curve fit, potency and CV, off-target selectivity, independent replication and claim honesty. It does not run an assay or prove in-vivo efficacy.

| Target | Binder | Verdict | Mean Kd | CV | Failed checks |
|---|---|---|---|---|---|
| 0 IL-6 | `GSTGRLQELARLLQGGP` | SUPPORTED | 12.0333 nM | 3.2066% |  |

## What this proves and what it does not
- Proves: the record is internally consistent, controls passed, the mean clears the potency threshold with acceptable replicate variation, off-targets clear the selectivity fold, and an independent lab repeated the protocol within tolerance.
- Does not prove: that the assay was run as described, that the binder works in an organism, or that the target is the right one. A fabricated record can look consistent.
- Verify offline: `python3 verify_offline.py run.json --pinned-key HkFG0ycWJ3ys2TSn51Ho+syxZsE6FWM7KL9DFqvZ+cQ=`