An AI-Driven Platform for In-Silico ADMET Prediction

ADMET-X is a comprehensive, AI-powered platform for predicting Absorption, Distribution, Metabolism, Excretion, and Toxicity (ADMET) properties of drug molecules based on their SMILES representations. It leverages machine learning models to provide accurate drug-likeness assessments and visual insights through radar plots and molecular drawings.

---

🚀 Project Overview

ADMET-X aims to accelerate early-stage drug discovery by providing in-silico predictions of pharmacokinetics and pharmacodynamics, reducing the need for extensive laboratory experiments.

Key highlights:

• Batch and single SMILES prediction.
• Downlodable report via CSV.
• Interactive radar plots for ADMET visualization.
• Toxicity prediction integrated with traditional ADME properties.
• Support for drawing molecules and uploading files.
• Ready-to-deploy with Docker and Fly.io.
• Used Vercel for FrontEnd deployment.

---

Graphical Abstract

---

📝 Architecture Diagram

---

💡 Features

• SMILES Input Options: Text input, file upload (.txt/.csv), drawing molecules, or example molecules.
• ADMET Predictions:
  • Absorption: Bioavailability, Caco2, HIA, etc.
  • Distribution: BBB, PPBR, etc.
  • Metabolism: CYP450 enzyme, etc.
  • Excretion: Clearance, Half-Life, etc.
  • Toxicity: AMES, Carcinogenicity, hERG, LD50, and more.
• Interactive Visualization: Molecule images, radar plots, and color-coded property status.
• Export Results: Download predictions as a CSV file.
• Local & Online Deployment: Works both locally and via Fly.io deployment.

---

📊 Datasets

• TDC(Therapeutics Data Commons): TDC - Used for Model -> training, testing and validating.
• PubChem: PubChem - Used for Batch Predictions.

---

🛠 Tech Stack

• Backend: Python, Flask, Joblib, Flask-CORS
• Frontend: Vite-React, TailwindCSS, Framer-Motion, Lottie
• AI/ML Models: Custom trained models for ADMET prediction
• Deployment: Docker, Fly.io and Vercel
• Utilities: RDKit, ChemUtils, Plotting modules

---

📁 Project Files & Navigation

ADMET-X/
├── BackEnd/                # Flask backend code, app.py, utils, Models folder
├── FrontEnd/               # React frontend code
├── Model_predictions/      # Predicted ADMET results (optional storage)
├── Model_training/         # Scripts and notebooks for training ML models
├── Test_Model/             # Unit tests or test scripts for models
├── admet_data/             # Raw datasets used for training/testing
├── LICENSE                 # MIT License file
├── README.md               # Project README with badges, instructions, contributors
├── SECURITY.md             # Security policy and responsible disclosure
├── example.py              # Example script demonstrating usage
├── package-lock.json       # Frontend dependency lock file
├── package.json            # Frontend dependency definitions
└── paths.py                # Paths configuration for project directories/files

---

⚙️ Installation (Local Setup)

1. Clone the repository

git clone https://github.com/yourusername/ADMET-X.git
cd ADMET/BackEnd

2. Setup Conda Environment

conda env create -f environment.yml
conda activate admet_env

3. Run BackEnd

python app.py

Running localhosts

Backend will run at: http://127.0.0.1:8080
Test endpoint: http://127.0.0.1:8080/ → should return "Backend is running."

4. Setup FrontEnd

cd ../FrontEnd
npm install
npm run dev

---

🌐 Deployment

1. Docker

docker build -t admet-ai .
docker run -p 8080:8080 admet-ai

2. Fly.io

"C:\Users\Rohith Reddy G K\.fly\bin\flyctl.exe" launch

• The BackEnd URL is: https://admet-backend.fly.dev

3. Vercel

• Deployed FrontEnd in Vercel.
• The FrontEnd URL is: https://admet-x.vercel.app

---

🖥 Usage

• Input molecule SMILES via text, file, draw, or example.
• Click Predict.
• View interactive ADMET radar plots and molecular images.
• Optionally, download all results as CSV.

---

👥 Contributors

Name	GitHub	LinkedIn
Sheik Arshad Ibrahim	GitHub	LinkedIn
Rohith Reddy G K	GitHub	LinkedIn
Sayed Jahangir Ali	GitHub	-
Thirumurugan M	GitHub	-

---

🧪 Contribution

Contributions are welcome! To contribute:

• Fork the repository
• Create a branch: git checkout -b feature-name
• Make changes and commit: git commit -m "Add new feature"
• Push to branch: git push origin feature-name
• Open a Pull Request

---

Curious to Cite ADMET-X?

General Citation Format:

Rohith Reddy.G.K, Sheik Arshad Ibrahim, et al. (2025). ADMET-X: AI-Driven Platform for In-Silico ADMET Prediction. [Online]. Available: https://admet-x.vercel.app

IEEE Format:

[1] Sheik Arshad Ibrahim, Rohith Reddy.G.K, et al., “ADMET-X: AI-Driven Platform for In-Silico ADMET Prediction,” 2025. [Online]. Available: https://admet-x.vercel.app

BibTex Format:

@software{admetx2025,
  author = {Sheik Arshad Ibrahim, Rohith Reddy.G.K and others},
  title = {ADMET-X: AI-Driven Platform for In-Silico ADMET Prediction},
  year = {2025},
  url = {https://admet-x.vercel.app},
  note = {Accessed: current_month current_year}
}

🌟If you like our project, give it a ⭐.