# Preparing your data

K is designed to work flexibly with the data clients provide. Within its supported modalities, K does not require data to conform to a single standard, ontology, or processing pipeline. However, the more standardized and homogeneous the data, the greater the analytical value K can deliver.

To help clients choose the right level of preparation, we define three tiers of data standardization, each with a different value-to-effort trade-off:

**Data Standardization Tiers**

* **Basic:** A small set of lightweight rules that the data must satisfy, for example, every clinical table must include a column with a unique patient identifier. These rules are minimal and easy to meet. The client is responsible for enforcing them. *(A full specification of these rules will be available soon and can be shared upon request.)*
* **Recommended:** An entry-level standardization effort focused on aligning key columns: demographics, procedural fields, and modality-specific measures (e.g., cell counts), to a shared data model. This makes the most common analytical operations directly comparable across datasets. Owkin can provide tooling (data contracts, methodology) and services to support these transformations.
* **Full:** Complete alignment to a common data model, ontology, and preparation pipeline. This level is typically required for use cases that involve cross-dataset cohorts or analyses run on merged datasets. Delivering this level of standardization requires a high-touch professional-services engagement from Owkin.

> **Optional enrichment:** Owkin has developed proprietary models that can extract additional features from raw data (for example, detecting cell types or adding bulk-deconvolution signatures) thereby augmenting the dataset before analysis.

**New modalities**

Adding aggregated tabular data for a new modality (such as a flow cytometry table) will not cause a technical failure in K. However, without a dedicated integration, the platform cannot guarantee the depth of analysis or reproducibility that a fully supported modality provides.

For this reason, we recommend clients to engage with the Owkin team before onboarding a new modality. This allows us to assess the data structure, confirm analytical coverage, and, where needed, implement the modality-specific logic required to deliver high-value, reproducible insights for the client's use case.