Interoperability points across products

[rrfreimuth]

GA4GH products utilize different technologies to support development of specifications and tooling. As a result, an instance of data might be serialized differently depending on the product with which it is used.

To improve the interoperability of GA4GH products with each other, key touch points between products should be identified and documented, and harmonized data structures should be developed when the same information is used by two or more products.

This issue proposes:

1. documenting practical use cases that utilize two or more GA4GH products (e.g., a workflow from Cloud WS or architecture from GIF)
2. developing an inventory of data elements that are common across two or more GA4GH products
3. identifying instances of data at touchpoints between products that are serialized differently
4. developing a process for harmonizing common data elements across GA4GH products
5. developing harmonized common data types and elements, and/or libraries of transformations, that improve interoperability across GA4GH products

[rrfreimuth]

See here for an example of preliminary work that has been done within the GKS WS (initially in 2020 and updated for VA-Spec in 2024).

Common data types could be categorized into 3 groups:

• Primitive types - lowest level of data representation, focused on what would be considered to be primitive types (e.g., numbers, strings, dates, etc)
• Generalized Domain types - reusable, minimal types that are used on the genomics domain and represent concepts that are common across several products (e.g., sequences, identifiers, ontology codes); built from primitives
• Object types - reusable, more complex types that are tightly scoped to a single concept (e.g., variation); built from generalized domain types and primitives, could be considered to be mini data schemas