Protean Labspeptide discovery infrastructureLive Base mainnet ledger
Autonomous peptide discovery is running live.
Protean is a scientific operating system for peptide discovery. It ingests evidence, generates peptide candidates, checks them through deterministic gates, prepares review-gated wet-lab handoffs, and publishes public records to the Protean Ledger on Base.
Models propose. Gates decide. The ledger keeps the receipts.
Live public artifact
Candidate inspection object
- Candidate
- cycle-20260526T020837Z-02-001
- Sequence visibility
- Approved public export
- Score band
- low
- Review gate
- awaiting_review
- Boundary
- Computational prioritization; not biological proof.
Inspect artifact (raw JSON)
{
"object": "candidate",
"id": "cycle-20260526T020837Z-02-001",
"sequence": "GHQMQHHCDDSQPTDCWP",
"sequence_visibility": "approved public export",
"score_band": "low",
"review_gate": "awaiting_review",
"public_fields": [
"sequence",
"lineage",
"structure_prior",
"claim_boundary"
],
"boundary": "computational prioritization; not biological proof"
}Sequences
15
Structures
15
Evidence sources
13
Digest
sha256:8b0d8070...a812
LIVE RUNTIME
Ledger records
240
Public sequences
15
Structures
15
Evidence sources
13
Lineage edges
30
Wet-lab packets prepared
1
Autonomous wet-lab submissions
0
Not a chatbot. Not a black box.
A scientific operating system,
not a single model output.
Evidence, sequence constraints, candidate families, structure priors, papers, and review records move through one typed runtime. Each stage leaves an inspectable trace so a reader can see how a candidate was proposed, why it was prioritized, and where the boundary of the claim sits.
Candidate intelligence
Sequence reasoning with context.
Protean connects full candidate sequences, lineage, structure priors, novelty checks, risk flags, and review state so prioritization is inspectable.
Surface: structures and candidate dossiers
Scientific memory
Failures become signal.
Evidence intake, contradictions, rejected motifs, and prior review outcomes accumulate as scientific memory instead of being erased between cycles.
Surface: research archive
Wet-lab handoff
Prepared packets, gated action.
Protean can prepare review artifacts for experimental follow-up. Approval here does not order assays, accept quotes, or claim validation.
Surface: docs and runtime boundary
Ledger provenance
Public records with lineage.
Approved research objects are published as typed records and lineage edges. The Ledger proves publication, integrity, and traceability, not biology.
Surface: Ledger and provenance graph
Receipts, not vibes.
Inspect the system
where it leaves evidence.
Protean's public site is not a brochure layer. It is an inspection surface for research artifacts, published sequences, structures, Ledger records, provenance paths, and technical documentation.
15 public candidates
Structures
Folded candidates, fold families, structural novelty, rerank context, and full sequences when publication-safe.
Open surface15 public artifacts
Research
Architecture paper, computational notes, mechanistic theses, and candidate dossiers with explicit claim boundaries.
Open surface240 records
Ledger
Canonical public scientific records on Base mainnet, with lineage edges and reproducible indexer Digest.
Open surfacePublic operations surface
Treasury
Inspect public treasury state, accounting surfaces, and protocol-facing operational context without treating finance as science.
Open surface30 lineage edges
Provenance
Candidate and family records expose sequences, hashes, lineage, and graph state without leaking private internals.
Open surfaceSystem documentation
Docs
Read the runtime architecture, discovery lifecycle, bounded learning, disclosure boundary, and scientific integrity rules.
Open surfaceMechanistic thesis layer
Papers
Daily computational proposals over trusted-tier evidence. These are research objects, not peer-reviewed findings.
Open surfaceOperational focus
Follow the candidate through the machine.
Protean moves peptide candidates from evidence ingestion to reviewed wet-lab handoff.
Current goal: identify and prepare the strongest candidates for wet-lab validation.
Snapshot - Jun 3, 5:46 AM - real artifact
Candidates
8
Active signals
6/7
Review holds
8
Wet-lab packets
1
Filters read candidate metadata, family lineage, and published sequence fields from reviewed artifacts.
Candidate progression trail
candidate sequence
GHQMQHHCDDSQPTDCWP
sha256:4158ed12...32b933
seqfam:4e57da74
rank 1
0.5436 dry-lab score
Conviction
What running Protean
teaches us.
The system's strongest claim is not that computation is truth. It is that scientific work becomes more useful when proposals, failures, reviews, and public records accumulate together.
01
Biology needs bounded autonomy: repeated work under explicit controls.
02
Provenance is not metadata; it is the product surface.
03
Assays are the truth layer. Computation is prioritization.
04
Sequences should be inspectable when publication-safe.
05
Scientific memory should accumulate across failures, not erase them.
Boundary
Review is a strength,
not a footnote.
Computational rankings are research-prioritization signals. Public records prove existence, lineage, authorship, and integrity of the research object. They do not prove biological activity, therapeutic effect, safety, efficacy, patentability, or clinical readiness.
01
No autonomous wet-lab submission.
02
No assay, activity, safety, efficacy, clinical, or therapeutic claims from computational ranking.
03
No scoring-weight, validator, prompt, or publication-policy mutation from the public site.
04
No private salts, credentials, embeddings, raw spectra, private manifests, or unreleased assay data.
05
No claim that on-chain anchoring proves biological truth.
Start inspecting
The live system is
the front door.
Researchers, labs, infrastructure partners, and skeptical readers can begin with the research archive, structures, Ledger, provenance graph, or docs. The public state is designed to be inspected before it is believed.