Capability Map

bijux-proteomics-knowledge preserves scientific assertions, their sources, their conflicts, and their biological context. It makes evidence reusable without pretending that storage, normalization, or trust scoring settles every scientific disagreement.

flowchart LR
    S[Sources] --> R[Evidence records]
    R --> C[Claims]
    C --> G[Integrity graph]
    G --> X[Conflict reconciliation]
    B[Biological grounding] --> C
    X --> D[Decision lineage and reviews]
    D --> K[Durable evidence memory]

Memory capabilities

Capability	Preserved context
Evidence records	kind, source, source type, origin, extraction method, assay context, quantitative support, artifacts, confidence, and time
Claims	statement, structured relation, assumptions, supporting and contradicting evidence, status, polarity, confidence, and resolution assays
Bundles	target-scoped evidence collections, document schema, decision tags, trust summaries, freshness, and conflict state
Integrity	valid claim-to-evidence links, graph coherence, duplicate and orphan detection
Reconciliation	policy, compared evidence, chosen action, rationale, belief update, actor, timestamp, and hold requirement
Query and lineage	structured filtering and traceability from decision areas to claims and evidence

Grounding capabilities

The package resolves protein identifiers, pathway and complex membership, kinase–substrate relationships, drug targets, disease terms, orthologs, and protein-feature overlaps. Every resolver carries status, ambiguity, coverage, or confidence rather than returning only matched values. Public reference workflows add citations, ontologies, literature audits, claim grounding, comparator confrontations, contradiction dossiers, and knowledge-deficit reports.

Boundary

Knowledge can describe support and conflict and propose a governed reconciliation. It cannot choose a portfolio action, authorize an experiment, or execute a workflow. Intelligence applies decision policy; lab governs experimental action; runtime governs execution.