Peptide RX

Peptide RX / feedback machine

Research memory that gets sharper with every proof spine.

The Learning Loop turns stored targets, folds, AlphaFold metrics, candidates, findings, timeline events, and Claim Graph debt into a public-safe feedback layer for the next research pass.

Memory is live. Automated model improvement requires the next outcome-label, retrieval-index, and calibration-dataset build.

Open Claim Graph Demo flow
42targets
38folds
2candidates
38findings
60events
234graph nodes

memory loop

01 Capture 82 02 Structure 38 03 Evidence 38 04 Debt 12 05 Feedback 40 06 Retrieval 234
feedback public-safe memory
01
Capture

Targets, folds, and candidates become stored research state.

active
02
Structure

Fold reports hold model confidence, coverage, and target context.

active
03
Evidence

Agent findings attach graded evidence and citations to objects.

active
04
Debt

Missing proof stays visible instead of being hidden.

active
05
Feedback

Events are the first ledger; explicit outcome labels come next.

partial
06
Retrieval

Graph nodes are ready to become searchable research memory.

next build

next best actions

What the system thinks should happen next.

high Attach stronger evidence or keep export limited PRX-GHSR-001

Candidate is grade D; public claims should remain bounded.

 high Attach stronger evidence or keep export limited PRX-GHSR-002

Candidate is grade D; public claims should remain bounded.

 medium Ingest functional-site annotations AGRP

Target context is missing residue-level functional-site support.

 medium Design or attach a candidate AGRP WT

The fold has no candidate workup connected yet.

 medium Ingest functional-site annotations BPC-157-target-cluster

Target context is missing residue-level functional-site support.

 medium Design or attach a candidate CAMP WT

The fold has no candidate workup connected yet.

 medium Ingest functional-site annotations CAMP-LL37

Target context is missing residue-level functional-site support.

 medium Ingest functional-site annotations CHRNA7

Target context is missing residue-level functional-site support.

 medium Design or attach a candidate CHRNA7 WT

The fold has no candidate workup connected yet.

 medium Design or attach a candidate CXCR4 WT

The fold has no candidate workup connected yet.

proof debt

What stays visible instead of being buried.

42 missing functional-site annotations ingest functional-site annotations TrkB / target 41 no candidate yet connect a candidate workup GLP1R WT / fold 40 computational-only status label outcome or add supporting proof PRX-GHSR-001 / candidate 40 low evidence grade attach graded evidence finding PRX-GHSR-001 / candidate 38 missing PAE/context where AlphaFold DB is not linked attach confidence metrics AlphaFold metrics / GHSR / alphafold metric 6 low-confidence AlphaFold regions likely attach confidence metrics LEPR WT / fold 6 no target linked yet label outcome or add supporting proof VEGFR2-tuned regenerative peptide with anti-fibrotic guard / idea 4 sequence redacted keep private sequence policy visible PRX-GHSR-001 / candidate

feedback ledger v0.1

Events are memory until explicit outcomes arrive.

observed studio_idea.created studio_idea
observed studio_idea.created studio_idea
observed studio_idea.created studio_idea
observed studio_idea.created studio_idea
observed studio_idea.created studio_idea
observed studio_idea.created studio_idea
observed studio_idea.created studio_idea
observed studio_idea.created studio_idea
observed studio_idea.created studio_idea
observed effect_query.matched effect_query
observed effect_query.created effect_query
supported effect_query.complete effect_query

What this is and is not

This is the first public learning-loop viewer. It proves Peptide RX is storing research memory and compiling feedback signals. It is not yet automated model retraining. The next build is explicit outcome labels, retrieval memory, and calibration data for future agents.