Project Themes:
- Automated Workflow Generation and Execution
Team Name: Rhea Workflow Team
Team Lead(s):
- Name: Chris Grams, Oleksandr Narykov
- Affiliation: Argonne National Laboratory, BV-BRC
- Email: cgrams@anl.gov, onarykov@anl.gov
Suggested Team Members and Roles [4-6 members]
| Name | Affiliation | Role / Expertise |
|---|---|---|
| Arvind Ramanathan | Argonne National Laboratory | AI Systems Integration Advisor |
| Chris Grams | Argonne National Laboratory, BV-BRC/ANL | Lead Developer, Galaxy Integration |
| Marius van den Beek | SCI-SCALE, BRC Analytics | Workflow Automation Engineer |
| Maulik Shukla | Argonne National Laboratory, BV-BRC | Use-case Alignment and Testing |
| Oleksandr Narykov | Argonne National Laboratory, BV-BRC/ANL | Workflow Automation Engineer |
Project Summary
This project extends the Rhea platform—an MCP+RAG-based environment that dynamically serves scientific tools—to enable automated workflow generation leveraging Galaxy and BV-BRC functionalities. By integrating BV-BRC’s pathogen data pipelines with Galaxy’s execution engine, the system will auto-generate end-to-end workflows for infectious disease analysis. This supports rapid hypothesis testing, reproducibility, and scalable execution of complex analyses across both ecosystems.
Goals and Objectives
- Goal 1: Integrate BV-BRC APIs and Galaxy tool registries into Rhea’s orchestration layer
- Goal 2: Implement LLM-driven workflow synthesis that identifies compatible BV-BRC and Galaxy modules
- Goal 3: Demonstrate automated workflow generation for at least two priority pathogens
Approach
Rhea will use retrieval-augmented generation (RAG) to identify relevant BV-BRC tools and Galaxy workflows via semantic search and metadata tagging. The system will then employ an LLM-based synthesis engine to auto-compose workflow specifications (e.g., CWL/YAML) executable within Galaxy. The team will test interoperability, runtime efficiency, and reproducibility using selected priority pathogens.
Data and Resources Required
| Resource Type | Source / Link | Description / Purpose |
|---|---|---|
| Data | BV-BRC genomes, priority pathogen datasets | Training and workflow inputs |
| Tools / Services | BV-BRC APIs, Galaxy server | Workflow integration and execution |
| LLMs / AI Models | Rhea MCP model, GPT-5 | Tool reasoning and synthesis |
| Compute / Storage | Argonne HPC / Galaxy cluster | Execution backend |
Expected Outcomes / Deliverables
- Integrated prototype showing end-to-end automated workflow synthesis
- Validation datasets
- Interoperability report for BV-BRC + Galaxy toolchains
Potential Impact and Next Steps
This project will provide a foundation for automated workflow assembly in bioinformatics, enhancing reproducibility and scalability for BV-BRC researchers. Follow-up activities may include connecting Rhea to HiPerRAG for data retrieval and integrating Co-Scientist reasoning modules for workflow evaluation.
Technical Support Needed
- GPU / LLM access
- API keys
- Mentor support
Additional Comments