The endpoint layer for peptide research.

0 · Abstract

The replication crisis in clinical research is a social-trust problem with a cryptographic solution. Peptide MD combines three primitives - a six-stage thesis pipeline (XTL), an outcome attestation (OEA), and a four-tier biosecurity gate (BioGate) - to produce a verifiable record from peptide signal to revealed cohort endpoint. We publish the verifier as @peptide-md/verifier; we publish three frozen test vectors; we publish ourselves through a contribution graph. The PDA-V1 design attestation is part of the same cryptographic chain, end-to-end.

1 · The Extended Thesis Loop

The Extended Thesis Loop (XTL) is the six-stage pipeline that runs every peptide candidate from signal intake through to a revealed cohort endpoint. Stages one through five carry the design discipline - citation-gated thesis generation with A/B/C/D/X evidence grading, where export is conditional on a Grade A or B citation. Stage six is the outcome layer that no in-silico oracle can produce:

1. 01Signal intake - eleven source connectors plus Oura/WHOOP/manual lab uploads
2. 02Novelty scoring - three-axis composite over PubMed, ClinicalTrials.gov, and the Peptide MD index
3. 03Thesis generation - citation-gated, three-scenario probability set
4. 04Scenario planning - buyer-safe, audience-typed dossier draft
5. 05Dossier export - seven-rule gate, evidence grade required at export
6. 06Endpoint evaluation - pre-registered cohort runs, OEA reveal, posterior update

2 · Outcome Evidence Attestation

An OEA-V1 record is a binding cryptographic commitment to (a) the primary endpoint, (b) the analysis plan, (c) the reveal window, and (d) the participant set. It is computed as:

oea_hash = SHA256( JCS(
  protocol_version  = "OEA-V1",
  cohort_id         = ...,
  primary_endpoint  = ...,
  analysis_plan_hash = SHA256(JCS(plan)),
  pre_registered_at = RFC3339,
  reveal_window     = [t0, t1],
  participant_root  = MerkleRoot(SHA256(JCS(p_i)) for p_i in participants),
  salt              = 32B random
))

At reveal time, the cohort host emits the unsalted analysis plan, the participant Merkle proofs, and the per-participant endpoint observations. Any third party can recompute oea_hash and compare it to the pre-registration. The protocol does not depend on a server; it depends on canonical JSON, SHA-256, and the RFC 6962 Merkle tree.

3 · BioGate

BioGate operates in two tiers. The public tier is open: every screening rule, profile-HMM, and cutoff is published to /architecture. The private tier requires advisory review before its parameters are committed; this asymmetry is deliberate. Public knowledge of every threshold is a biosecurity risk, but private knowledge of the entire pipeline is a transparency risk. We publish the shape of the private tier (rule count, calibration metric, audit schedule) without publishing the cutoffs.

4 · Living Outcome Oracle

The LOO maintains a Beta(α, β) posterior per (peptide, indication, baseline biomarker bin, cohort cohort) tuple. Successful primary-endpoint outcomes, weighted by evidence grade and BioGate tier, increment α; null or adverse outcomes increment β. Reputation is a math object, not a brand asset; an OEA reveal updates the posterior and exposes the inverse cumulative as a calibrated probability for any downstream cohort. Posteriors are surfaced on every peptide page, every protocol card, and every dossier export.

5 · External attestation interop

A peptide entry can carry zero or more PDA-V1 design attestations alongside its OEA-V1 record. The two protocols are deliberately disjoint in their cryptographic claims - PDA binds a peptide design to its inputs; OEA binds a human-use cohort to its endpoint. External attestations from third-party design pipelines are accepted via the same verifyPDA entry point, marked with source: "external", and surface the same audience-posture rules as Peptide MD-issued attestations.

6 · Tokenomics (preliminary)

IPNFTs follow Molecule's V2 specification. Each IPNFT routes through a royalty cascade that splits incoming revenues across (a) cohort treasury, (b) cohort participants via the cooperative, (c) originating researchers, and (d) cohort stakers. A native token $PMD exists only as a placeholder governance primitive; mainnet issuance is conditioned on jurisdictional review and is not part of the v1 launch.

7 · Disclosures

• Peptide MD is research infrastructure. Nothing on this site is medical advice. Consult a licensed clinician before any human use.
• The mock data and deterministic hashes shown across this site are illustrative until v1 launch; the verifier and the canonical-JSON dialect are not mocks.
• BioGate is calibrated against published toxin and select-agent registries. Private-tier thresholds are reviewed by an external biosecurity advisor before any change.
• Token issuance is contingent on jurisdictional review. No representation of expected return is made or implied.
• Source repository, audit reports, and the contribution graph are public. Security disclosures: security@peptide-md.io.