Inocras Reveals Design for Optimizing Clinical Genomics Pipeline Operations on AWS HealthOmics

Inocras migrated the infrastructure of its whole-genome cancer analysis solution to AWS HealthOmics, reducing analysis costs by 72% and improving processing speed by 47%. This material introduces the architecture and operational automation strategies that made these results reproducible on a daily basis.

Architecture and Operational Automation

The data flow is configured so that EventBridge rules trigger Lambda to start HealthOmics workflows (StartRun) upon S3 uploads. The complex four-stage chaining (Secondary→Tertiary→Annotation→QC) uses EventBridge rules and Lambda for state tracking and branching logic, while DLQ and Archive features prevent event loss. Deployment follows a GitOps approach integrating Bitbucket, AWS CodeConnections, and CodePipeline, minimizing manual intervention and supporting version-specific rollbacks and audit trails (chain of custody).

Compliance and Security Controls

To comply with certifications such as HIPAA and HITRUST-CSF, KMS encryption, network isolation via VPC Endpoint (PrivateLink), and access control based on AWS SSO were applied. Patient identification information is replaced with anonymous sample_id, and mapping information is managed only within access-controlled LIMS. CloudTrail Data Events and HealthOmics Data Provenance features are combined to automatically log all data access and execution history.

Cost Optimization Strategy

To prevent over-provisioning, the open-source CLI HealthOmics Run Analyzer is utilized. Run Status Change events trigger analysis tasks on ECS Fargate, and the generated CPU/memory recommendations are reflected in the WDL runtime, repeating a right-sizing cycle. This enables quantitative data-driven resource optimization with every workflow version update.