RFC: Enhance AWS HealthOmics MCP Server with Data Store Management Capabilities

Summary

Propose enhancing the existing aws-healthomics-mcp-server with comprehensive data store management capabilities to create a complete genomic research solution covering both workflow management and data operations.

Problem Statement

The current AWS HealthOmics MCP server provides excellent workflow management capabilities (create, execute, analyze WDL/CWL/Nextflow workflows) but lacks data access and management functionality. Genomic researchers need both capabilities for complete workflows:

1. Data Management: Access sequence stores, variant stores, reference stores, annotation stores
2. Data Import: Seamless local file → S3 → HealthOmics import workflows
3. Data Analysis: Search variants, fetch sequences, get coverage profiles
4. Access Control: IAM role management and validation

This creates a gap where researchers must use separate tools or write custom scripts for data operations while using the MCP server for workflow management.

Proposed Solution

Enhance the existing aws-healthomics-mcp-server by adding complementary data store management tools, creating a comprehensive solution that covers:

Current Capabilities (Workflow Management)

• ✅ Create and version workflows
• ✅ Execute and monitor workflow runs
• ✅ Performance analysis and troubleshooting
• ✅ Workflow linting and validation

Proposed New Capabilities (Data Management)

• 🆕 Sequence Store Operations: List, search, and retrieve genomic sequences
• 🆕 Variant Store Operations: Search variants by gene, position, impact, frequency
• 🆕 Reference Store Operations: Manage and access reference genomes
• 🆕 Annotation Store Operations: Search annotations and submit VCF files
• 🆕 Data Import Workflows: Local file → S3 → HealthOmics orchestration
• 🆕 S3 Integration: Discover genomic files, manage bucket access

Implementation Approach

New Tool Modules

Following the existing server structure, add these new tool modules:

awslabs/aws_healthomics_mcp_server/tools/
├── sequence_store_tools.py     # Sequence data operations
├── variant_store_tools.py      # Variant search and analysis
├── reference_store_tools.py    # Reference genome management  
├── annotation_store_tools.py   # Annotation operations
└── data_import_tools.py        # S3 integration and import

Design Principles

• Non-breaking: Preserve all existing functionality
• Consistent: Follow existing patterns and conventions
• Comprehensive: Add full test coverage
• Documented: Update README and documentation

Tool Integration

The new tools will integrate seamlessly with existing workflow tools:

• Import data using new tools → Execute workflows using existing tools
• Analyze results using both data access and workflow analysis tools

Value Proposition

For Genomic Researchers

• Complete Solution: One MCP server for all HealthOmics operations
• Streamlined Workflows: No need for separate tools or custom scripts
• Natural Language Interface: AI assistants can orchestrate complex multi-step operations

For the AWS MCP Ecosystem

• Enhanced Adoption: More comprehensive HealthOmics coverage
• Community Value: Addresses real user needs in genomic research
• Reference Implementation: Demonstrates best practices for complex AWS service integration

Example Use Cases

Complete Research Workflow

User: "Import my local FASTQ files, run my variant calling workflow, and analyze the results"

AI Assistant orchestrates:
1. [NEW] Upload files to S3 using data_import_tools
2. [NEW] Import to HealthOmics using sequence_store_tools  
3. [EXISTING] Execute workflow using workflow_execution
4. [NEW] Analyze variants using variant_store_tools
5. [EXISTING] Generate performance report using run_analysis

Multi-Store Data Discovery

User: "What genomic data do I have available across all my HealthOmics stores?"

AI Assistant provides:
- [NEW] Sequence store inventory
- [NEW] Variant store summaries
- [NEW] Reference genome catalog
- [NEW] Available annotations

Implementation Details

Backward Compatibility

• Zero breaking changes to existing functionality
• All existing tools and APIs remain unchanged
• New tools are additive enhancements

Code Quality Standards

• Follow existing code patterns and structure
• Comprehensive unit and integration tests
• Type hints and documentation
• Pass all pre-commit hooks

Documentation Updates

• Expand main README with new capabilities
• Add tool documentation following existing format
• Update configuration examples
• Create usage examples for common workflows

Open Questions

1. Tool Naming Convention: Should we follow the existing AHO prefix pattern (e.g., ListAHOSequences) or use a different convention for data tools?

2. Configuration: Should data store IDs be configured via environment variables (like workflow execution) or discovered dynamically?

3. Error Handling: Should we extend the existing error handling patterns or create new ones for data operations?

Request for Feedback

This RFC proposes a significant enhancement that would make the AWS HealthOmics MCP server the definitive solution for genomic research workflows. The implementation:

• ✅ Fills a real gap in current functionality
• ✅ Follows AWS MCP patterns and guidelines
• ✅ Maintains backward compatibility
• ✅ Adds significant value for users

We would appreciate feedback on:

• Overall approach and feasibility
• Implementation strategy and structure
• Any concerns or alternative approaches
• Timeline and coordination preferences

References

• Existing AWS HealthOmics MCP Server
• AWS HealthOmics Documentation
• Model Context Protocol Specification

Proposed by: Peter (CarlsbergGBS)
Implementation Ready: Full working implementation available for review
Timeline: Ready to start immediately upon approval