What Is Vertical Interoperability in Biomedical Research?
Vertical interoperability is the property of research data that remains understandable, reusable, and verifiable across every stage of its lifecycle — from collection through analysis, publication, and reuse — because the data, code, and workflows travel together with standardised, persistent, machine-readable metadata. Where traditional (horizontal) interoperability lets two systems exchange data at a single point in time, vertical interoperability keeps meaning, provenance, and governance attached to the data as it moves through time and between research stages.
Why vertical interoperability matters now
Biomedical research is producing more data than ever, yet a persistent complaint from researchers and funders alike is that more data has not translated into proportionally more actionable knowledge. A large part of the reason is structural: the research data lifecycle has historically been treated as a linear pipeline — design a study, collect data, analyse it, publish, deposit the dataset, move on. Every handoff in that pipeline loses information. The analysis code is separated from the dataset. The consent terms are separated from the derived variables. The harmonisation decisions live in a lab notebook nobody can find three years later.
Policy has now caught up with the problem. The NIH Data Management and Sharing Policy, in effect since January 2023, requires funded researchers to plan for data sharing from the outset — and the wider funding ecosystem is converging on the view that deposit alone is not the goal. Data that is deposited without aligned metadata, runnable code, and traceable provenance is technically shared but practically unusable. Vertical interoperability names the missing property: shared data should arrive with everything a stranger needs to understand, trust, and reuse it, and that packaging should be created during research, not reconstructed afterwards.
Horizontal vs vertical interoperability
The distinction is easiest to see side by side.
| Dimension | Horizontal interoperability | Vertical interoperability |
|---|---|---|
| What it connects | Two systems at the same lifecycle stage (e.g., two EHRs exchanging a record) | All stages of one dataset’s lifecycle (collection → analysis → publication → reuse) |
| Core mechanism | Exchange standards such as HL7 FHIR messages between live systems | Self-describing packages: data + code + workflow + machine-readable metadata that stay aligned |
| Typical failure | Systems cannot talk to each other | Systems talk, but meaning and provenance are lost at each handoff |
| Who benefits first | Clinicians and operational teams | Researchers reusing data they did not collect — and the funders who paid for it |
| Representative standards | HL7 FHIR, DICOM, X12 | RO-Crate, FAIR principles, schema.org-based metadata, persistent identifiers |
The two are complements, not rivals. A hospital network can be excellent at horizontal exchange and still produce research datasets that are vertically opaque — harmonisation that happens once, undocumented, at analysis time is precisely the information loss vertical interoperability is designed to prevent.
The building blocks of vertical interoperability
Self-describing data and code packages
The packaging layer is where the concept becomes concrete. Community standards such as RO-Crate wrap a dataset, its analysis code, its workflow definitions, and its metadata into a single machine-readable research object. A package that describes itself can be validated, indexed, and reused without an email to the original author — which is the practical test of whether sharing has actually happened.
Metadata captured natively, not retrospectively
Retrospective FAIRification — cleaning up metadata months after the work, to satisfy a compliance requirement — is the administrative tax that makes researchers resent data-sharing mandates. The emerging alternative is AI-assisted metadata capture inside the working environment itself: agents and tooling that record variable definitions, transformations, and provenance as a by-product of doing the analysis. When the metadata is generated where the work happens, alignment across lifecycle stages stops being a separate chore.
Ethical provenance that travels with the data
Consent terms, de-identification decisions, and population-representation caveats are metadata too — arguably the most important kind. Vertical interoperability extends to machine-readable ethical provenance, so that a dataset’s permissible uses remain attached to it through every derivation. A secondary researcher should be able to ask not only “what does this variable mean?” but “what am I allowed to do with it?” and get a machine-checkable answer.
Workflows as first-class, registered artefacts
Reproducibility requires that the computation itself be shareable. Registries of reusable, containerised workflows — built on engines such as Nextflow — turn a paper’s methods section from prose into something executable. When the workflow is a citable, versioned artefact, “we followed the published pipeline” becomes a verifiable claim rather than an aspiration.
The institutional barrier is bigger than the technical one
A consistent finding across the research-infrastructure community is that the primary obstacle to vertical interoperability is institutional, not technical. The standards exist; the packaging formats exist; the compute exists. What is scarce is organisational readiness: incentives that reward well-packaged data, training that embeds these practices into existing workflows rather than adding a parallel bureaucracy, and funding models that sustain infrastructure beyond a single grant cycle. Any serious adoption plan pairs the compliance “stick” with an immediate workflow “carrot” — tooling that makes the researcher’s own next analysis easier, with the interoperable package produced as a side effect.
Where federation fits
Vertical interoperability and federated analysis solve complementary halves of the same problem. Sensitive health data often cannot be centralised at all — which is why federated Trusted Research Environments bring the analysis to the data, and data never leaves the source. But federation without vertical interoperability produces siloed answers: each site computes locally against its own undocumented conventions. Federation with vertical interoperability is a different proposition — standardised, self-describing data at every node, registered workflows that run identically across sites, and provenance and airlock-governed outputs that document exactly what was computed, where, and under whose authority. In a federated network, the metadata layer is not a nice-to-have; it is the only thing that makes cross-site results interpretable.
Practical steps for research organisations
Organisations that want to get ahead of this shift do not need to wait for mandates. A pragmatic sequence: adopt a packaging standard (RO-Crate is the current centre of gravity) for new studies; standardise on a common data model such as OMOP for observational data, so that metadata alignment is a mapping problem rather than a translation problem; move analysis into workflow engines with versioned, containerised pipelines; and instrument the research environment to capture provenance automatically. Institutions running Trusted Research Environments have a head start — the TRE is the natural place to enforce packaging, provenance, and governance without adding burden to individual researchers.
Frequently asked questions
What is vertical interoperability in simple terms?
It means a dataset stays understandable and reusable through its whole life, because its meaning, code, and permissions travel with it — instead of being lost each time the data changes hands or stages.
How is vertical interoperability different from FAIR?
The FAIR principles (Findable, Accessible, Interoperable, Reusable) describe desirable properties of shared data. Vertical interoperability describes how those properties are achieved and maintained across lifecycle stages: through self-describing packages, native metadata capture, and provenance that survives every handoff. FAIR is the outcome; vertical interoperability is the mechanism.
What is RO-Crate?
RO-Crate (Research Object Crate) is a community standard for packaging research data together with its metadata, code, and context in a machine-readable form, built on schema.org vocabulary. It is the leading candidate for the “self-describing package” layer of vertical interoperability.
Does vertical interoperability require centralising data?
No. It is arguably most valuable in federated settings, where data stays with its custodian. Standardised metadata and registered workflows are what make results computed at different sites comparable — without any pooling of record-level data.
Why do data-sharing policies alone not solve this?
Policies can mandate deposit, but deposited data without aligned metadata, runnable code, and traceable provenance is shared in name only. The bottleneck is the quality and persistence of the packaging, which policy can encourage but tooling must deliver.
Who should own vertical interoperability in a research organisation?
In practice it lands with whoever operates the research computing environment — because the environment is where metadata can be captured natively. Data stewards define the standards; the platform enforces them as a by-product of normal work.