Molecular Toxicity Prediction Project

Overview

This project focuses on developing a machine learning application for predicting molecular toxicity. By utilizing advanced deep learning models (Artificial Neural Networks) and Bayesian optimization, our solution surpasses existing tools like MolToxPred and ToxiM. We built a custom dataset from public databases such as T3DB and research papers, applying feature engineering to gain enhanced data insights.

Features

• Custom Dataset: Compiled from public databases (T3DB) and research papers.
• Feature Engineering: Applied to extract meaningful insights and improve model performance.
• Model Evaluation: Compared multiple models including XGBoost, Logistic Regression, and ANN to identify the best-performing model.

Data Collection

• Sources: Public databases like T3DB and various research papers.
• Preprocessing: Data cleaning, normalization, and feature extraction were performed to prepare the dataset for model training.

Models and Techniques

Artificial Neural Networks (ANN)

• Deep learning model chosen for its superior performance in this domain.
• Implemented with Bayesian optimization for hyperparameter tuning.

Evaluation Metrics

• Accuracy: Percentage of correctly predicted instances.
• Precision: Proportion of true positive predictions among all positive predictions.
• Recall: Proportion of true positive predictions among all actual positives.
• F1 Score: Harmonic mean of precision and recall, providing a balance between the two.

Results

• ANN with Bayesian Optimization: Outperformed other models, showing the highest accuracy and best overall performance.
• Comparison: ANN > XGBoost > Logistic Regression in terms of accuracy and predictive capability.

Installation

1. Clone the repository:

   git clone https://github.com/ANGADJEET/MolToxInsight

Running the Project

1. Navigate to the neuralTox directory:

   cd neuralTox

Backend

1. Navigate to the backend directory:

   cd backend

2. Run the server:

   python app.py

3. Wait for the server to start.

Frontend

1. Navigate to the frontend directory:

   cd ..
   cd frontend

2. If running for the first time or the node modules folder is not there

   npm install

   npm install axios

   npm install react-router-dom

3. Run the development server:

   npm run dev

4. Click on the provided link to access the application.

Acknowledgements

• T3DB: For providing comprehensive data on toxic compounds.
• Research Papers: For the foundational knowledge and additional data sources.
• Open Source Libraries: Including Scikit-learn, TensorFlow, and XGBoost, which made this project possible.

For any questions or issues, please open an issue on the repository or contact us at angadjeet22071@iiitd.ac.in arav22091@iiitd.ac.in.