AWS HealthOmics

AWS HealthOmics is not a general-purpose batch service with a genomics label. It is a purpose-built service for bioinformatics workflows and omics data handling, with separate primitives for running workflows and for storing sequence, variant, and annotation data.

That split is the key design boundary. Workflows manage computation, while the omics stores manage durable research artifacts. Architectures that treat those as one thing usually lose provenance, access control clarity, or cost control.

As of March 10, 2026, AWS documents that variant stores and annotation stores are not open to new customers, although existing customers can continue to use them. That changes the teaching arc: new omics programs should think of HealthOmics first as a workflow and sequence-data platform, then make an explicit plan for how long-lived derived data and interpretation assets will be stored and governed.

AWS documents private workflows as user-created workflows written in WDL, CWL, or Nextflow. Around those definitions, HealthOmics adds the operational controls that research and production programs actually need: immutable workflow versions, run groups to control compute consumption, and call caching to avoid unnecessary recomputation.

The StartRun API makes those controls concrete. Operators decide which workflow version is being invoked, which IAM role executes it, which run group governs compute, where outputs land, and how much temporary storage is allocated for the run. That turns workflow execution into an explicit operations surface instead of a black-box batch job.

HealthOmics separates sequence storage from variant and annotation analysis for a reason. Sequence stores keep genomics reads and related files. Variant stores hold called variants for analytics. Annotation stores add contextual knowledge that is imported against the same coordinate system as the variant data.