Peptiter / DiscoveryLab
Workflow
Product workflow

Product workflow — Program, Discover, Evidence, Portfolio, Handoff.

The five-workspace progress ribbon that organizes a candidate's journey through the shipped product.

Product workflow

From therapeutic intent to testable, mechanism-checked peptide families.

The shipped app presents a five-workspace progress ribbon as the primary navigation. Every step produces a reviewable artifact, including pathway claims, Lean verification receipts, lineage, and unresolved evidence gaps.

Program
GLP1R · Class B GPCR · seed exendin-4
Program

identity panel, progress ribbon, next-best action

Discover

BioScout rail, working set, selected lineage

Evidence

in-silico bundle, claim ledger, Lean audit drawer

Portfolio

Pareto front, active-learning batch, evidence gaps

Handoff

export manifest, LabSpace batch, returned results

  1. Step 01: Select condition

    Define therapeutic intent or phenotype as the entry point.

    in app: Program ▸ setup
    Condition diagram for Obesity (ICD-10 E66)Central condition node labelled E66 surrounded by four annotation labels: phenotype, intent, comorbid, and contraindications.ICD-10 E66 (Overweight and obesity); MeSH D009765. Anchored to standard vocabularies.E66Phenotype: observable clinical traits (HPO terms) used to constrain the search.phenotypeTherapeutic intent: agonism, antagonism, allosteric modulation, or biased signalling.intentComorbidities: co-occurring conditions that adjust target prioritisation and selectivity requirements.comorbidContraindications: receptor or pathway interactions to avoid (off-target, safety liabilities).contra-ind.
  2. Step 02: Map pathways & receptors

    Pathways, receptor families, ligand classes, candidate mechanism claims, contraindications.

    in app: Program ▸ target
    Pathway and receptor graph for ObesityDirected graph from condition vertex through pathway nodes (incretin, GLP-1 / cAMP, insulin secretion) to four receptor candidates and verifiable mechanism claims: GLP1R, GIPR, GCGR, Y2R.Pathway graph edges: weighted by curated evidence, graph-AI hypotheses, ligand-receptor curation, and later checked by the mechanism verifier.Condition vertex · ObesityUpstream pathway · incretinDirect mechanism · GLP-1 / cAMPDownstream effector · insulin secretionReceptor candidate · GLP1RGLP1RReceptor candidate · GIPRGIPRReceptor candidate · GCGRGCGRReceptor candidate · Y2RY2RConvergence node: receptors with overlapping endogenous ligands.Obesitycond.Pathway layer: intermediate biological processes linking condition to druggable targets.pathwayReceptor layer · primary: GLP1R · Class B GPCRreceptor
  3. Step 03: BioScout source systems

    Evidence-backed mimicry plans from evolved peptide systems.

    in app: Discover ▸ BioScout
    BioScout source systems diagramThree source organism rows (Gila monster, amphibian, fish gut) feeding a curated motif library labelled GLP-1 motif.Heloderma exendin-4 lineage: long-acting GLP-1R agonist scaffold.Gila monsterFrog skin peptide families with insulinotropic activity.amphibianTeleost proglucagon-derived peptides as alternative scaffolds.fish gutMotif library · GLP-1 motif: curated sequence motifs and pharmacophores indexed from APD3, DRAMP, ConoServer.GLP-1 motif
  4. Step 04: Seeded evolution

    Start from known bioactive peptide families and evolve local analogs with ancestry.

    in app: Discover ▸ lineage
    Seeded evolution ancestry tree starting from exendin-4Phylogenetic-style tree from seed peptide exendin-4 branching into analogs of class DL-GLP-A1, with operator history preserved on each branch.Ancestry edges: each branch records the operator applied and the family constraint preserved.Seed · exendin-4: validated parent peptide with known bioactivity.Seed · exendin-4: validated parent peptide with known bioactivity.Branch operator: substitution / cyclisation / N-methylation under family constraints.Branch operator: substitution / cyclisation / N-methylation under family constraints.Analog leaf · DL-GLP-A1-class candidate with full operator history.Analog leaf · DL-GLP-A1-class candidate with full operator history.Analog leaf · DL-GLP-A1-class candidate with full operator history.Analog leaf · DL-GLP-A1-class candidate with full operator history.Seed peptide · exendin-4exendin-4Analog · DL-GLP-A1DL-GLP-A1
  5. Step 05: Visualize peptide–receptor fit

    Structure-aware 3D review of binding orientation.

    in app: Evidence ▸ structure
    Peptide–receptor fit diagram for GLP1R · Class B GPCRReceptor scaffold of GLP1R · Class B GPCR with peptide backbone Cα trace of DL-GLP-A1 placed against the pocket; key contact at ECD contact.Receptor scaffold · GLP1R · Class B GPCRPeptide backbone · DL-GLP-A1 Cα trace placed against the receptor pocket.Cα atom · residue 1 of DL-GLP-A1Cα atom · residue 2 of DL-GLP-A1Cα atom · residue 3 of DL-GLP-A1Cα atom · residue 4 of DL-GLP-A1Cα atom · residue 5 of DL-GLP-A1Cα atom · residue 6 of DL-GLP-A1Key contact · ECD contact (predicted distance < 4 Å).ECD contact
  6. Step 06: In-silico + mechanism verification

    Multi-criteria scoring, pathway reachability checks, Lean audit artifacts, perturbation evidence, and rejection gates pre-wet-lab.

    in app: Evidence ▸ audit
    In-silico rejection gates for DL-GLP candidatesSix horizontal score bars (fit, stab, sol, aggr, synth, tox) with pass/fail labels for each gate.Receptor fit: composite score from docking pose quality and contact-residue agreement. · score 0.91 · passes gate thresholdfitpassStability: predicted resistance to proteolysis and conformational entropy penalty. · score 0.78 · passes gate thresholdstabpassSolubility: CamSol / SolubiS-style intrinsic solubility proxy. · score 0.84 · passes gate thresholdsolpassAggregation propensity: Zyggregator / Tango-style β-aggregation score. · score 0.66 · passes gate thresholdaggrpassSynthetic feasibility: SPPS coupling-difficulty estimate plus length and modification penalties. · score 0.74 · passes gate thresholdsynthpassToxicity prior: ToxinPred-class classifier; failing candidates are gated out. · score 0.38 · passes gate thresholdtoxpass
  7. Step 07: Prepare wet-lab batch

    Hand off through LabSpace to capability-matched partners.

    in app: Handoff ▸ batch
    Wet-lab batch manifest for DL-GLP candidatesManifest table of five candidate IDs with prefix DL-GLP and four assay vials whose heights encode activity readout (EC50 · cAMP assay).Batch manifest: machine-readable handoff (SiLA 2 / Allotrope ADF) with sequences, modifications, and assay plan.Candidate DL-GLP-0421: passes in-silico gates, queued for SPPS synthesis and binding assay.DL-GLP-0421Candidate DL-GLP-0438: passes in-silico gates, queued for SPPS synthesis and binding assay.DL-GLP-0438Candidate DL-GLP-0455: passes in-silico gates, queued for SPPS synthesis and binding assay.DL-GLP-0455Candidate DL-GLP-0461: passes in-silico gates, queued for SPPS synthesis and binding assay.DL-GLP-0461Candidate DL-GLP-0473: passes in-silico gates, queued for SPPS synthesis and binding assay.DL-GLP-0473Vial 1: EC50 · cAMP assay readout — height encodes activity.Vial 2: EC50 · cAMP assay readout — height encodes activity.Vial 3: EC50 · cAMP assay readout — height encodes activity.Vial 4: EC50 · cAMP assay readout — height encodes activity.
  8. Step 08: Receive results & refine

    Wet-lab and perturbation feedback update ranking, mechanism assumptions, and next-assay selection.

    in app: Portfolio ▸ feedback
    Wet-lab feedback closed loop diagramClosed loop between a wet-lab node returning EC50 · cAMP assay measurements and a re-rank node that updates the surrogate model and selects the next batch.Closed loop: wet-lab measurements update the surrogate model and re-rank candidates (Bayesian / active learning).Wet-lab node · EC50 · cAMP assay returned through LabSpace.EC50Re-rank node: surrogate model updated with new evidence; acquisition function selects next batch.re-rank

Tip · Focus a step with Tab, move between steps with /, then Tab into the diagram to read each element.

Discovery Strategy spec
Full technical write-up of the workflow blocks, diagram elements, pathway-verification layer, and references.