MyTxGNN Project Introduction
MyTxGNN is a drug repurposing prediction platform for Malaysia, built on Harvard's TxGNN model published in Nature Medicine. It provides AI-powered predictions for potential new therapeutic uses of NPRA-approved drugs.
Project Overview
| Metric | Value |
|---|---|
| Drugs Analyzed | 508 |
| KG Predictions | 41,560 |
| DL Predictions | 9,968,985 |
| High Confidence (≥0.7) | 176,021 |
| Unique Diseases | 17,041 |
Key Features
1. Dual Prediction Methods
- Knowledge Graph (KG): Fast queries based on existing drug-disease relationships
- Deep Learning (DL): Neural network-based predictions with confidence scores
2. FHIR R4 API
Access predictions programmatically using the FHIR R4 standard:
curl https://mytxgnn.yao.care/fhir/MedicationKnowledge/db00860.json
3. Evidence Levels
Each prediction is assigned an evidence level (L1-L5) based on supporting clinical evidence.
4. Malaysia Focus
Focused on drugs registered with Malaysia’s National Pharmaceutical Regulatory Agency (NPRA), ensuring relevance to local healthcare needs.
Data Sources
| Source | Description | Records |
|---|---|---|
| NPRA Registration | Malaysia pharmaceutical products via data.gov.my | 27,938 |
| TxGNN | Harvard knowledge graph (Nature Medicine 2023) | 17,080 nodes |
| DrugBank | Drug standardization and mapping | 508 matched |
Technology Stack
- Backend: Python, TxGNN, DGL (Deep Graph Library)
- Frontend: Jekyll, GitHub Pages
- Standards: FHIR R4, SMART on FHIR, CDS Hooks
- Data: TxGNN Knowledge Graph, DrugBank, NPRA
Use Cases
Research
- Identify drug repurposing candidates for rare diseases
- Explore drug-disease relationships in the knowledge graph
- Access prediction data via API for downstream analysis
Clinical Decision Support
- Integrate with EHR systems via SMART on FHIR
- Receive alerts for potential new indications
- Access evidence summaries for repurposing candidates
Education
- Learn about drug repurposing methodologies
- Understand knowledge graph approaches
- Explore FHIR/SMART standards implementation
Team
MyTxGNN is developed and maintained by the yao.care research team, building on the foundational work of Harvard’s Zitnik Lab.
Citation
If you use this dataset or software, please cite:
@misc{mytxgnn2026,
title={MyTxGNN: Drug Repurposing Predictions for NPRA-Approved Drugs in Malaysia},
url={https://mytxgnn.yao.care},
year={2026}
}
And cite the original TxGNN paper:
@article{huang2023txgnn,
title={A foundation model for clinician-centered drug repurposing},
author={Huang, Kexin and Chandak, Payal and Wang, Qianwen and Haber, Shreyas and Zitnik, Marinka},
journal={Nature Medicine},
year={2023},
doi={10.1038/s41591-023-02233-x}
}
Related Resources
- TxGNN Paper - Nature Medicine 2023
- TxGNN GitHub
- TxGNN Explorer - Interactive prediction queries
- NPRA Malaysia - National Pharmaceutical Regulatory Agency
Disclaimer
This platform is for research purposes only and does not constitute medical advice. Drug repurposing predictions require clinical validation before application.
Last updated: 2026-03-03 | MyTxGNN Research Team
This platform is for research purposes only and does not constitute medical advice. Drug repurposing predictions require clinical validation before application.
Last updated: 2026-03-03 | MyTxGNN Research Team