SEC · SCIENTIFIC CURATION · EVIDENCE MANAGEMENT
Sec is being developed to support source selection, metadata normalization, and evidence review. The interface is available; the runtime is planned. Automated services and connectors require implementation and testing. The proposed publication workflow requires human review and links claims to methods and sources.
01 · PROPOSED CAPABILITIES
The proposed workflow covers collection, normalization, validation, interpretation, review, and publication. Each stage should record inputs, decisions, versions, and responsible parties to distinguish observed data, inferences, hypotheses, and recommendations. The cards describe planned tasks; they do not confirm deployed services.
Record primary and secondary sources with identifiers, authorship, date, method, version, scope, and conditions of use.
Standardize and combine datasets to compare entities and identify associations without treating them as causal relations.
Analyze relations among taxa, genes, proteins, and metabolites with methods from systematics, phylogeny, and comparative genomics.
Record the origin, transformations, relations, and versions of data between source and publication.
Prepare pages, dashboards, charts, and structured packages for databases or APIs from reviewed content.
Challenge extrapolations, compare sources, and state when a conclusion depends on inference.
Select authoritative sources and record why each source is relevant to the question.
Reconcile names, identifiers, units, and controlled vocabularies across heterogeneous records.
Distinguish experimental evidence, review, inference, hypothesis, and recommendation.
Resolve accepted names, synonyms, taxon scope, and classification before biological comparison.
Prepare alignments, infer trees, and record support values, methods, and analytical uncertainty.
Compare genes, proteins, orthologs, and genomic regions while preserving version context.
Compare studies and produce citation-linked summaries while preserving differences in method, sample, and population.
Record hypotheses, assumptions, predictions, and tests as distinct scientific objects.
Represent entities, relations, and provenance in schemas, matrices, knowledge graphs, and API records.
Check whether outputs cite sources, answer the question, follow the defined format, and can be reproduced.
Apply controls to prevent secrets and private context from entering public products by default.
Record runs, inputs, versions, decisions, and publication state to support auditing and reproduction.
Present reviewed results with audience-appropriate language, citations, limitations, and a clear visual hierarchy.
Define a reproducible sequence from question and source selection through analysis, review, and delivery.
02 · WORKFLOW
For requests requiring different specialties, the planned workflow would define stages, select agents, check their results, and assemble a deliverable for human review.
Define the question, organism, domain, time scale, and decision the result should support.
Activate curation, evolution, data, security, product, or QA agents according to the task.
Compare responses, identify conflicts, check cited evidence, and record uncertainty.
Prepare the response, record, interface, or data package with documented sources, versions, and decisions.
GOVERNING RULE
The workflow should record information sources and distinguish observed data, interpretations, hypotheses, and recommendations.
03 · SOURCES AND CONNECTORS
The proposed connectors would query sources, standardize retrieved records, and preserve identifiers. Each integration must report its deployment status and tested version, and respect access permissions, usage rights, rate limits, and attribution rules. Listing a source does not confirm an active connector.
Locate articles, reviews, metadata, authors, identifiers, and bibliographic relations.
source_idpublication_datemethod_scopeCheck accepted names, synonyms, authors, lineages, and taxonomic identifiers.
taxon_idaccepted_nameclassificationQuery sequences, assemblies, genes, and genomic regions while preserving the record identifier and version.
accessionsequence_scopeversionConnect proteins, orthology, annotated function, experimental evidence, and evolutionary comparison.
protein_idevidence_codeorthology_relationRelate structures, ligands, residues, conformational states, and predicted models with their limitations.
structure_idexperimental_methodmodel_statusOrganize compounds, bioactivity, and health indicators without mixing molecule, product, market, and population.
compound_idassay_contextpopulation_scopeResolve DOI metadata, abstracts, full-text links, publication status, and citation context.
doiabstract_statusfull_text_routeConnect raw reads, projects, samples, and study context while respecting controlled-access boundaries.
run_accessionsample_idaccess_policyRelate protein domains, families, conserved regions, and functional interpretation.
domain_idfamily_idannotation_evidenceMap pathways, reactions, organism context, and molecular processes to biological entities.
pathway_idreaction_idorganism_scopeConnect protein associations while keeping experimental, curated, co-expression, and predicted channels distinct.
protein_pairevidence_channelconfidence_scoreQuery series, samples, platforms, study designs, and information about data processing.
series_idsample_idplatformOrganize occurrence records, coordinates, observation uncertainty, and taxonomic context.
occurrence_idgeospatial_scopeobservation_qualitySeparate institutional indicators, regulatory records, population scope, and policy interpretation.
indicator_idjurisdictioninterpretive_boundaryConnect accessions, germplasm records, material context, and collection metadata.
accession_idmaterial_typecollection_context04 · PROPOSED ARCHITECTURE
The proposal separates scientific rules, the installable package, source adapters, and consuming projects. This division allows each component to be versioned and limits the data and functions available to each project.
The planned core would contain capabilities, contracts, policies, vocabularies, and records that should remain consistent across Sec versions.
{ skills, contracts, policies, registry }The planned package would define product metadata, interface, capability status, version, and agent activation conditions.
plugin.manifest(capabilities, interface, version)The proposed adapters would convert data from Kannabium, plant_connectors_core, and bibliographic services into the contracts defined by the core.
adapter.connect(source, contract, policy)Each project should receive only approved functions and contexts. Access to private data must be explicit, limited, and recorded.
project.consume(plugin, approved_context)ACADEMIC AND COMMERCIAL APPLICATION
In research projects, the planned workflow would link each result to its source, method, version, uncertainty, and applied taxonomic identification. The four proposed layers would separate the scientific core from distributed components, with rules for licensing, integration, access, and new domains. Access to private context would require explicit project authorization.
05 · INTEGRATIONS
The planned integration would receive a question, query the required sources, and return data or analyses for review. Reference pages remain responsible for presenting approved content.
06 · VISUAL ASSETS
The images show the Sec mascot in different Kannabium contexts. They are illustrations for institutional communication and do not represent observations, experiments, or scientific results.
