Following recent Clinical & Phenotypic Data Capture (Clin/Pheno) Work Stream
Phenopackets calls, we are moving into an active phase of infrastructure updates and deep interoperability with major health data standards, framed by your contributions on real-world implementation. As we have successfully done in the past, and to ensure these modifications are practical, sustainable, and reflective of diverse research practices and goals, we are starting a community consultation.

We are seeking detailed requirements, implementation feedback, and critiques from developers, clinical geneticists, data engineers, and consortia on the open discussion topics outlined below.

Areas awaiting your input

The proposed updates have been categorised into a structured project plan spanning technical baseline fixes, schema refinements, and external interoperability layers:

1. Protobuf and architectural infrastructure

• Upgrading serialisation baseline: We are exploring an upgrade to the latest versions of Protocol Buffers (such as moving from our current proto3 baseline to Protobuf Editions). This change would re-enable explicit optional field labels, directly addressing several community ticket requests. We need feedback on how these adjustments might impact downstream versioning, reference implementations, and your existing local builds.

• Migration strategy: As we pave the way for a transition from Phenopackets v2.0 to v3.0, the working group has aligned on a development pipeline. We seek input on what automated transformation tooling, validation libraries (across Java, Python, and Rust), and quality-control checks your organisation would need to migrate existing datasets safely.

2. Examples of proposed clinical modelling enhancements

• Episodic and recurring events: Conditions such as seizures occur multiple times over a lifespan. Because the current schema only captures a single Onset and Resolution per feature, we are evaluating a new field tentatively called Time_Observed or Age_Observed to robustly capture longitudinal disease progression.

• Demographic variables (Gender, Sex, and Ethnicity): Handling these variables is highly complex across regions and clinical domains. Moving forward, we are proposing a strict structural separation between biological/genetic features (e.g. sex assigned at birth, chromosomal characteristics) and broader socio-administrative contexts (gender identity, ancestry, etc.). We need feedback on how your systems capture this data and which universally accepted ontologies (such as HANCESTRO or AFPO) should be prioritised.

• Decoupling variant representation: For fields of study such as pharmacogenomics, variants of interest are often identified without an associated disease manifestation. Our working group is reviewing proposals to allow standalone variant interpretations outside of a locked diagnostic case hierarchy.

• Survival time validation (Cardinality vs. Serialisation): Implementers have reported that certain language libraries (particularly in Python) unintentionally serialise missing integer values as zero. We are consulting the community on a proposal to update the schema/software to throw an absolute error if survival_time is explicitly set to 0 to prevent skewed survival data.

3. Interoperability & Export Pipelines

• FHIR, OMOP, and Beacon ecosystems: Phenopackets rarely originate in a vacuum; they must be seamlessly extracted from Electronic Health Records (EHRs) and large observational databases. We welcome developers currently building automated ETL extraction tools or concept-mapping pipelines to help standardise our export workflows across Phenopackets, HL7 FHIR, and the OMOP Common Data Model.

How to share your feedback

To ensure transparency and centralised issue tracking, we have migrated these discussions into GitHub tickets.

1. Join the upcoming Phenopackets Community Calls, which will occur on:

   • July 23 at 7 am PT (check your local time here)

   • August 13 at 8 am PT (check your local time here)

   • August 27 at 7 am PT

   • September 10 at 8 am PT

   • If you don’t already have a calendar invitation, please reach out to Beatrice Amos (clinpheno_wsm@ga4gh.org) to receive one.

2. Review the topics: A list of actively discussed topics is found in our Project Plan document (with links to GH tickets).

3. Comment on GitHub: Please visit our repository’s issue tracker to share your thoughts, edge cases, and votes directly on open tickets (linked in the document above).

We recognise that the ultimate success of Phenopackets relies on a collaborative, implementer-led approach to standard maturity. We thank you in advance for your time and expertise as we build a more connected, AI-ready future for precision medicine.