- Helps with
- Findability, accessibility, interoperability, and reusability of data, metadata, and infrastructure.
- Best for
- Use as the outcome framework and assessment lens across the full R&D lifecycle.
- Watch out
- Principles describe desired behavior, not a single technical architecture or conformance test.
Profile library
Browse the knowledge base
Find the standard that fits the work in front of you—not just the acronym you already know.
54 of 54 profiles · Select up to three to compare.
- Helps with
- Protocol-to-analysis clinical and nonclinical research data, including acquisition, tabulation, analysis, and submission structures.
- Best for
- Best for regulated studies and traceable submission packages; less natural for early discovery or raw instrument output.
- Watch out
- Conformance is detailed and version-sensitive; transformations can preserve structure while losing source context.
Evidence & profile details
- Helps with
- Modular resources, profiles, terminology bindings, and APIs for electronic healthcare data exchange.
- Best for
- Operational exchange at system boundaries, eSource acquisition, registries, and clinical-research integrations.
- Watch out
- Base FHIR conformance does not imply conformance to a named research IG; local profiles can diverge, and FHIR is not an analysis-ready warehouse.
Evidence & profile details
- Helps with
- Relational structure, conventions, and standardized vocabularies for longitudinal observational health data.
- Best for
- Multi-source cohort analytics, patient-level prediction, characterization, and network studies after ETL.
- Watch out
- ETL is expensive, source nuance can be compressed, and vocabulary maintenance is an ongoing operational dependency.
Evidence & profile details
- Helps with
- Analytical instrument data, contextual metadata, semantic models, ontologies, and long-lived data packaging.
- Best for
- Vendor-neutral lab data acquisition, instrument integration, archive, and cross-technique reuse.
- Watch out
- ADF/ADM and ASM do not share one access path; technique coverage, converter behavior, model versions, and extension governance must be pinned. AFO openness does not make the whole stack open.
Evidence & profile details
- Helps with
- Experimental design, sample characteristics, protocols, assay technologies, and sample-to-data relationships.
- Best for
- Multi-omics and multi-assay study metadata at the boundary between experiment design and repository submission.
- Watch out
- Rich metadata entry is labor-intensive and local templates can drift without governance and validation.
Evidence & profile details
- Helps with
- A family of interoperable biological and biomedical ontologies governed by principles for openness, scope, identifiers, relations, and maintenance.
- Best for
- Semantic annotation, knowledge graphs, terminology normalization, and cross-dataset integration.
- Watch out
- Coverage and maintenance vary by ontology; overlap, versioning, and term-selection policy still require local stewardship.
Evidence & profile details
- Helps with
- An OWL 2 ontology for interoperable provenance using entities, activities, agents, and qualified relationships.
- Best for
- Cross-system lineage, transformation history, audit evidence, and knowledge-graph provenance.
- Watch out
- The model is intentionally generic; useful provenance requires a scoped profile, identifier policy, and capture instrumentation.
Evidence & profile details
- Helps with
- Life-science profiles over Schema.org for datasets, tools, workflows, samples, proteins, genes, and related Web resources; status varies by profile.
- Best for
- Web-scale discovery and lightweight metadata publication alongside richer repository records.
- Watch out
- Profiles have different release states; markup improves discovery but is not a substitute for a domain data model.
Evidence & profile details
- Helps with
- A JSON-LD metadata document that aggregates data, code, workflows, people, instruments, and contextual entities into a research object.
- Best for
- Portable dataset packages, workflow exchange, preservation, publication, and handoff between repositories and compute environments.
- Watch out
- Core conformance is deliberately lightweight; interoperability depends on shared profiles and validation beyond the base crate.
Evidence & profile details
- Helps with
- An RDF vocabulary for interoperable catalogs of datasets, data services, distributions, dataset series, versions, and qualified relations.
- Best for
- Enterprise and federated data catalogs, cross-repository discovery, and standardized catalog APIs.
- Watch out
- DCAT describes catalog resources, not the internal scientific schema; useful deployment needs a domain profile and controlled vocabularies.
Evidence & profile details
- Helps with
- Composable APIs for identifying and retrieving data, submitting workflows, and enabling federated genomic analysis.
- Best for
- Cloud and federated genomics where data access and computation must work across heterogeneous repositories.
- Watch out
- The APIs solve infrastructure interoperability, not dataset semantics, consent harmonization, or analytical comparability on their own.
Evidence & profile details
- Helps with
- Machine-readable dataset metadata, resources, record structure, ML semantics, provenance, and usage-policy extensions.
- Best for
- The final mile from governed data product to portable, loadable ML dataset across tools and repositories.
- Watch out
- A newer cross-domain standard; life-science conventions and BioCroissant profiles are still developing.
Evidence & profile details
- Helps with
- Persistent and resolvable identifiers, predictable metadata retrieval, and typing for machine-actionable digital objects.
- Best for
- Long-horizon infrastructure design for object-level interoperability across repositories and automated agents.
- Watch out
- The available documentation is explicitly incomplete and should not be treated as a comprehensive final specification.
Evidence & profile details
- Helps with
- Standards-agnostic biomedical concept definitions plus implementation artifacts such as SDTM Dataset Specializations and value-level metadata.
- Best for
- Computable clinical concepts that connect protocol, collection design, tabulation metadata, and downstream automation.
- Watch out
- Content is informative and incrementally curated; concept, specialization, terminology, and downstream standard versions must be pinned together.
Evidence & profile details
- Helps with
- Medical imaging information objects, encoding, media, services, workflow, security, terminology resources, and DICOMweb.
- Best for
- Clinical imaging acquisition, archive, exchange, and imaging-derived research datasets.
- Watch out
- Conformance is feature-specific; private tags, de-identification, modality variation, and AI cohort labels require explicit profiles and tests.
Evidence & profile details
- Helps with
- Identification of medicinal products, pharmaceutical products, substances, dose forms, routes, units, and packages across the product lifecycle.
- Best for
- Regulated medicinal-product master data, cross-system product identity, and jurisdictional submissions such as EMA SPOR/PMS.
- Watch out
- The suite spans multiple ISO standards and amendments; implementation scope, identifiers, and timelines vary by jurisdiction and are still evolving.
Evidence & profile details
- Helps with
- Human- and machine-readable case-level phenotypic, clinical, diagnosis, measurement, biosample, and genomic interpretation data.
- Best for
- Portable phenotype/genotype exchange for rare disease, cancer, registries, diagnostics, and computational analysis.
- Watch out
- Flexible optionality and ontology dependence require application-specific validation; it does not replace an EHR API, consent layer, or cohort warehouse.
Evidence & profile details
- Helps with
- Cloud/object-store bioimaging data in Zarr v3 with axes, multiscales, transforms, labels, and high-content-screening plates.
- Best for
- Large multidimensional microscopy, high-content screening, multiscale visualization, and cloud-native image analysis.
- Watch out
- Pre-1.0 changes and transitional metadata remain; writer/viewer compatibility and round-trip preservation must be tested with the chosen toolchain.
Evidence & profile details
- Helps with
- Vendor-neutral mass-spectrometry spectra plus acquisition, instrument, and processing metadata using controlled vocabulary terms.
- Best for
- Raw-to-open conversion and exchange of MS spectra across proteomics and metabolomics toolchains.
- Watch out
- Conversion can lose vendor-specific detail; XML is large, and mzML does not capture the full cross-sample design, identifications, or quantification results.
Evidence & profile details
- Helps with
- RDF terms for quality dimensions, metrics, measurements, policies, certificates, and annotations.
- Best for
- Publishing the evidence behind data-quality claims in catalogs, knowledge graphs, and governed dataset releases.
- Watch out
- DQV does not define universal quality metrics or decide fitness for use; projects must define and justify their own measurements and thresholds.
Evidence & profile details
- Helps with
- Standardized biomedical data-use permission terms for matching controlled-access datasets to research purposes.
- Best for
- Consent-aware discovery, data access review, and machine-readable permitted-use conditions in genomics and health research.
- Watch out
- Ontology matching cannot resolve jurisdiction, contract, consent nuance, expiry, or downstream duties without authoritative policy and human governance.
Evidence & profile details
- Helps with
- Shapes for validating RDF graphs against structural and semantic constraints, with machine-readable validation reports.
- Best for
- Executable conformance checks for linked-data metadata, profiles, and knowledge graphs.
- Watch out
- Passing shapes proves only the encoded constraints; it does not prove scientific truth, completeness, ontology fitness, or relational table quality.
Evidence & profile details
- Helps with
- Govern, Map, Measure, and Manage functions for addressing AI risks across organizations and system lifecycles.
- Best for
- Governance overlay for intended use, accountability, risk measurement, release decisions, and ongoing monitoring.
- Watch out
- Voluntary and use-case agnostic; it does not prescribe life-science schemas, legal compliance, or quantitative acceptance thresholds.
Evidence & profile details
StandardSAM/BAM · CRAM · VCF/BCF
GA4GH HTS Format Specifications
AcquireHarmonizeExchangeLearn + reuse
- Helps with
- Sequence alignments, compressed reference-oriented reads, variant calls, binary encodings, standard tags, and associated indexes.
- Best for
- The base interchange layer for sequencing alignments and variant-call datasets across pipelines, archives, and analysis tools.
- Watch out
- Format conformance does not establish sample identity, consent, reference correctness, variant normalization, QC, or pipeline reproducibility. FASTQ has no formal hts-specs definition.
Evidence & profile details
- Helps with
- Tabular sample-to-data relationships, biological and technical factors, replicates, instruments, acquisition context, and proteomics experimental design.
- Best for
- Proteomics studies that need an explicit, machine-readable mapping from biosamples and factors to raw and processed mass-spectrometry files.
- Watch out
- It does not encode downstream statistical-analysis parameters or results. Working-branch templates and rules can move ahead of the final PSI specification, so release and validator versions must be pinned.
Evidence & profile details
- Helps with
- Dataset layout, filenames, participants, sessions, acquisition metadata, events, coordinate systems, and derivatives across MRI, PET, EEG, MEG, iEEG, microscopy, and related modalities.
- Best for
- Human- and machine-readable packaging of neuroimaging and behavioral studies for validation, sharing, and reproducible analysis.
- Watch out
- Passing BIDS validation proves encoded structure, not image quality, biological plausibility, complete metadata, de-identification, or analysis validity. Draft BEPs are not released specification content.
Evidence & profile details
- Helps with
- Neurophysiology acquisition and processed data, time series, events, stimuli, behavior, electrophysiology, optical physiology, devices, and experimental metadata.
- Best for
- Session-level packaging and reuse of complex neurophysiology experiments where synchronized signals and experiment context must remain together.
- Watch out
- Extensions and optional fields can fragment interoperability; storage and API compatibility must be tested, and schema validity does not establish signal quality or biological correctness.
Evidence & profile details
- Helps with
- Mass-spectrometry identification results, search inputs, peptide-spectrum matches, peptides, proteins, scores, thresholds, and crosslinking support.
- Best for
- Detailed, tool-independent exchange of proteomics identification evidence for post-processing, validation, archiving, and repository submission.
- Watch out
- The XML model is complex; older versions remain in operational use, and producer/consumer support varies by feature. It does not encode the full sample design or statistical analysis.
Evidence & profile details
- Helps with
- Minimum information for interpretable and reproducible microarray and high-throughput sequencing studies, including design, samples, raw and processed data, and protocols.
- Best for
- A submission and publication completeness gate for functional-genomics studies, especially when preparing repository records and supporting data.
- Watch out
- These are minimum-information checklists rather than one executable schema. Stewardship is legacy, and newer assay classes such as single-cell data require current repository guidance and additional profiles.
Evidence & profile details
- Helps with
- Tab-delimited summaries of mass-spectrometry-derived proteins, peptides, spectra, small molecules, features, identifications, and quantitative values.
- Best for
- Accessible, computational result exchange when consumers need a concise table rather than the complete identification or quantification evidence model.
- Watch out
- The two branches are not interchangeable, and a summary cannot reconstruct the full processing history or detailed evidence. Tool support must be checked against the exact flavor and version.
Evidence & profile details
- Helps with
- Computational biological model structure and mathematics through SBML, plus simulation setup, model changes, algorithms, tasks, outputs, and data references through SED-ML.
- Best for
- Reproducible exchange and execution of systems-biology models and simulation experiments across compatible tools.
- Watch out
- SBML package and tool support varies, and a syntactically reproducible simulation does not prove biological validity, parameter identifiability, or agreement with experimental evidence.
Evidence & profile details
- Helps with
- Vendor-neutral XML structures for analytical samples, experiment steps, methods, result series, audit trails, digital signatures, and technique-specific definitions.
- Best for
- Open analytical-instrument exchange and archival where a supported AnIML technique definition captures the required method and result semantics.
- Watch out
- The official schema remains Draft 0.90. Technique coverage, current conformance tooling, vendor support, and independent production adoption are unverified and must not be implied.
Evidence & profile details
- Helps with
- XML exchange of experimental information about compound synthesis, reactions, products, procedures, measurements, and compound testing.
- Best for
- Cross-system exchange of medicinal-chemistry synthesis and testing records that are not covered by instrument-only or generic study metadata.
- Watch out
- The public repository has no packaged releases and showed no verified maintenance after 2021. Current adoption, extension governance, validator support, and successor plans are unverified.
Evidence & profile details
- Helps with
- Reusable indicators, priorities, maturity levels, and evaluation guidance for assessing data and metadata against the FAIR principles.
- Best for
- Use as the evidence rubric that turns FAIR from an aspiration into a repeatable release and improvement assessment.
- Watch out
- It is not a certification, and locally adapted scoring or weighting means totals from different assessment tools are not automatically comparable.
Evidence & profile details
- Helps with
- Identifiers, creators and contributors, titles, publisher, dates, resource types, versions, rights, funding, subjects, geolocation, and typed relations among research outputs.
- Best for
- Canonical publication metadata for datasets and related software, workflows, projects, instruments, and publications receiving DataCite DOIs.
- Watch out
- It describes and cites a research output but does not specify its internal scientific schema, validate quality, or enforce access and reuse conditions.
Evidence & profile details
- Helps with
- Machine-readable concepts for data and processing, purposes, legal bases, parties, recipients, rights, risks, controls, technologies, AI, and jurisdiction-specific laws.
- Best for
- Privacy and data-protection context for human, genomic, clinical, real-world, and AI datasets where DUO alone is too narrow.
- Watch out
- DPV is a vocabulary, not legal advice or an enforcement engine; local authority, consent, contracts, and jurisdiction-specific interpretation remain controlling.
Evidence & profile details
- Helps with
- Policies containing permissions, prohibitions, duties, parties, assets, constraints, inheritance, and conflict strategies.
- Best for
- Machine-readable usage conditions for datasets and distributions, including research-purpose, redistribution, attribution, retention, and temporal or jurisdictional constraints.
- Watch out
- A syntactically valid policy does not prove the assigner has authority, make the policy legally enforceable, or provide the system that evaluates and enforces it.
Evidence & profile details
- Helps with
- A system bill of materials spanning software, builds, AI models, datasets, identities, provenance, integrity, licenses, security findings, and relationships.
- Best for
- Reproducible and reviewable supply-chain records for an AI release that combines scientific data, preprocessing code, dependencies, models, and licenses.
- Watch out
- It is a broad BOM model rather than a scientific metadata profile; generated inventories require verification, and license metadata does not authorize use of sensitive human data.
Evidence & profile details
- Helps with
- A JSON descriptor for a coherent collection of resources, plus Data Resource, Table Schema, and Table Dialect specifications.
- Best for
- Lightweight packaging and validation of assay exports, reference tables, tabular analysis results, and other file-based data products.
- Watch out
- It does not supply biological semantics, full provenance, privacy policy, repository trust, or ML-specific intended-use and bias documentation.
Evidence & profile details
StandardISO/IEC 5259
ISO/IEC 5259 Data Quality for Analytics and Machine Learning
PlanAcquireHarmonizeExchangeLearn + reuse
- Helps with
- Terminology and examples, data-quality measures, management requirements, a process framework, and a governance framework for analytics and ML data.
- Best for
- The quality-management spine for training, validation, and evaluation data used in life-science analytics and ML.
- Watch out
- The normative publications are not freely available, the series is cross-domain, and it does not provide life-science thresholds, domain semantics, or regulatory approval.
Evidence & profile details
- Helps with
- Structured, audience-aware summaries of dataset origins, collection and annotation, intended use, evaluation context, ethical considerations, and decisions affecting downstream performance.
- Best for
- Human-facing readiness and release documentation for clinical, imaging, omics, laboratory, and real-world ML datasets.
- Watch out
- There is no single mandatory schema or conformance test; narrative claims require linked evidence, ownership, review, and update controls.
Evidence & profile details
- Helps with
- Runtime and design-time events for jobs, runs, and datasets, with extensible facets for source code, schemas, versions, quality metrics, assertions, and other lineage context.
- Best for
- Operational lineage instrumentation across ETL, ELT, laboratory, feature-engineering, and model-data pipelines.
- Watch out
- Lineage is only as complete as its instrumentation; inconsistent naming, missing events, facet-version drift, and backend retention can leave an incomplete history.
Evidence & profile details
Validation standardCoreTrustSeal
CoreTrustSeal Trustworthy Data Repositories Requirements
ExchangeLearn + reuse
- Helps with
- Repository organizational infrastructure, digital-object management, technology, security, designated-community service, continuity, curation, and long-term preservation.
- Best for
- Repository selection and assurance where life-science data must remain authentic, understandable, accessible, and reusable over time.
- Watch out
- Certification concerns a repository and its declared scope, not the scientific quality, ethics, or AI fitness of every deposited dataset; certification is time-bound.
Evidence & profile details
- Helps with
- A computable study definition spanning objectives, endpoints, eligibility, interventions, schedule of activities, amendments, estimands, and protocol content.
- Best for
- Upstream protocol facts that must flow consistently into study-build, registry, document, and downstream data systems.
- Watch out
- USDM is a model and reference architecture—not an EDC, submission dataset, or proof that generated documents comply with every regulator—and every linked terminology and API version must be pinned.
Evidence & profile details
- Helps with
- Vendor-neutral exchange and archival of study metadata, subject data, administrative data, reference data, and audit information.
- Best for
- EDC, eClinical, archive, and study-operation transfers where the operational study record—not only submission tables—must remain portable.
- Watch out
- ODM v2 is not backward-compatible in several areas, XML is the currently supplied schema serialization, and schema validity does not prove semantic or regulatory fitness.
Evidence & profile details
TerminologyLOINC
Logical Observation Identifiers Names and Codes
AcquireHarmonizeExchangeLearn + reuse
- Helps with
- Identifiers for laboratory tests, clinical observations, survey instruments, panels, and answer lists.
- Best for
- Normalizing what was measured or observed across laboratories, EHRs, FHIR, CDISC, and real-world-data models.
- Watch out
- A LOINC code does not by itself preserve local method nuance, result quality, unit correctness, or mapping confidence, and implementations must track deprecated and replacement concepts.
Evidence & profile details
- Helps with
- Polyhierarchical clinical concepts, descriptions, relationships, reference sets, and compositional semantics for detailed healthcare meaning.
- Best for
- Conditions, findings, procedures, anatomy, organisms, and other clinical concepts requiring more semantic depth than flat billing classifications.
- Watch out
- Licensing and distribution vary by territory, national extensions can diverge, and post-coordination support is not uniform across systems.
Evidence & profile details
- Helps with
- A machine-processable syntax and semantics for units, prefixes, compound units, and conversions.
- Best for
- Quantitative values that must survive movement across instruments, laboratories, FHIR resources, CDISC datasets, and analytical stores.
- Watch out
- Syntactic validity does not prove that a unit is clinically appropriate, that a numerical value is plausible, or that arbitrary units are mutually convertible.
Evidence & profile details
- Helps with
- Hierarchical coding of adverse events, medical history, indications, investigations, product issues, and related regulatory medical concepts.
- Best for
- Clinical-trial safety coding, individual case safety reports, aggregate safety analyses, and regulatory pharmacovigilance exchange.
- Watch out
- MedDRA is licensed, version-sensitive, and multiaxial; a coded term does not establish seriousness, expectedness, relatedness, or causality.
Evidence & profile details
- Helps with
- Medicinal-product names, ingredients, countries, marketing authorization holders, strengths, ATC classifications, and drug groupings for medication coding.
- Best for
- Concomitant medication, prior therapy, exposure, and pharmacovigilance drug coding across global clinical programs.
- Watch out
- Dictionary data require a subscription, ambiguous product names still need expert review, B3/C3 choices affect granularity, and up-versioning can change records and classifications.
Evidence & profile details
- Helps with
- A relational representation of EHR, claims, prescribing, laboratory, patient-reported, and related data for distributed patient-centered research.
- Best for
- Analyses that must run consistently across PCORnet partners while each institution retains operational control of local data.
- Watch out
- The CDM deliberately preserves source values and does not itself impose all plausibility or consistency edits; ETL fidelity and study-specific fitness remain separate gates.
Evidence & profile details
- Helps with
- Nineteen linked tables supporting standardized queries over enrollment, encounters, diagnoses, procedures, dispensing, prescribing, laboratory, death, patient-reported, and feature-engineering data.
- Best for
- FDA-style distributed active surveillance of drugs, biologics, devices, and vaccines using partner-held healthcare data.
- Watch out
- The public repository’s latest tag is v8.2.2 while the landing page still names v8.1.0 and v8.2.0 concurrently; the current partner deployment mix is not publicly verified.
Evidence & profile details
Metadata profileVulcan RWD IG
HL7 Vulcan Retrieval of Real World Data for Clinical Research
AcquireExchange
- Helps with
- FHIR profiles and queries for finding cohorts and retrieving a minimal EHR-derived dataset for retrospective clinical research and potential regulatory use.
- Best for
- Research applications retrieving RWD from FHIR-capable EHRs through a named, testable research profile rather than unconstrained base resources.
- Watch out
- STU1 is limited to retrospective EHR RWD; prospective eSource, registries, payer data, downstream transformation, and analysis readiness are outside current scope.