Cross-Species Aging Knowledge Integration into Agentic AI Platform Uncovers Conserved Mechanisms - Supplementary

1. EvoAge_neo4J.zip

This directory contains the single-file neo4j.dump backup of the full EvoAge Knowledge Graph. This file is the complete, archived snapshot of the 1.04 billion-triple graph, enabling any researcher to load and rebuild the exact EvoAge graph instance for full reproducibility and querying of all raw data and processed relationships.

2. Final_trained_model.zip

This directory contains the final, production-ready Knowledge Graph Embedding (KGE) models essential for the EvoAge platform's predictive capabilities. The RotatE model (128-dimensional) trained on the full EvoAge graph is the production model used in the EvoAge chatbot backend for real-time link prediction and hypothesis testing. The RESCAL model (512-dimensional) is the top-performing embedding from the supplementary "Aging" benchmark dataset. Each sub-directory includes the entity (*.npy) and relation (*.npy) embeddings, along with the necessary config.json for immediate loading and use in prediction tasks.

3. used_in_backend.zip

This directory contains all the files used in backend as an artifacts of trained model and other mapping files required by EvoAge chatbot.

4. KG_files.zip

This directory contains the Knowledge Graph (KG) data for EvoAge. It includes the tensor file (.pt) of the 1-hot encoded triples, along with the necessary node and relation mapping files (.pkl and .txt) to interpret the data.