Repository navigation
This release archives the source code for miRKatAI, the AI-powered querying agent presented in Chapter 3 of the PhD thesis: "Computational approaches for therapeutic target discovery to ameliorate muscle wasting during ageing" (University of Galway, 2025).
Key Functionality:
- Architecture: A multi-agent system built using LangGraph and Python.
- LLM Integration: Orchestrates Google Gemini models (Flash 2.0/2.5) for natural language understanding and reasoning.
- Agent Modules:
- Master Router: Decomposes complex user queries.
- SQL Agent: Converts natural language to MySQL queries for miRKatDB.
- Literature Agent: Performs RAG-based searches on peer-reviewed scientific literature.
- Plotting Agent: Generates data visualisations using Python (Matplotlib/Seaborn).
Thesis Context:
This code corresponds to the methodology described in Section 3.2.2 ("miRKatAI: An AI-Powered Querying and Analysis Agent") and the results presented in Section 3.3.2.