AIDrugApp: Auto-Multi-ML Module

AIDrugApp is an AI-powered virtual screening web-app for drug discovery. This repository contains the code for the Auto-Multi-ML module, which allows users to develop and compare multiple machine learning models to select the best-performing algorithm for molecular data. Additionally, the app helps predict target data based on user-specific machine learning models.

Features

• Auto-Multi-ML Module: Aids in developing and comparing various machine learning models for classification and regression tasks, helping users select the best-performing model for their data.
• Exploratory Data Analysis (EDA): Provides detailed insights into the uploaded and feature-engineered data.
• Multiple Machine Learning Models: Performance comparison of variety of ML algorithms (e.g., Random Forest, Logistic Regression, Decision Tree).
• Data Prediction: Enables users to predict target data using their trained models.
• Downloadable Reports and Data: Model performance and predictions can be downloaded as CSV files.

Installation

1. Install Python and necessary libraries:

   pip install streamlit pandas scikit-learn sweetviz lazypredict numpy

2. Clone the repository:

   git clone https://github.com/DivyaKarade/CustomML.git
   cd AIDrugApp

3. Run the application:

   streamlit run app.py

Instructions for the Auto-Multi-ML Tool

Sidebar Menu

• Auto-Multi-ML Checkbox: Enables the Auto-Multi-ML module for model comparison and selection.
• Upload .csv Files: Upload data files for building models and making predictions.
  • Example input files are provided in the repository: Example .csv input file for ML model building Example .csv input file for predictions

Step-by-Step Instructions

1. Select Algorithm: Choose between 'Classification' or 'Regression'.
2. Upload Descriptor Data: Upload a .csv file containing descriptor data (with target data for model building).
3. Select Exploratory Data Analysis (EDA) Options: Check the boxes for EDA on uploaded or feature-engineered data.
4. Build ML Model: Evaluate multiple models and select the top-performing one for predictions from options such as Random Forest, Logistic Regression, Decision Tree, etc.
5. Upload New Data for Prediction: Provide data excluding target values to predict using the selected model.
6. View and Download Results: Results are displayed, and users can download predictions.

Output

• Performance metrics of different ML models.
• Predictions based on the selected model.
• Downloadable CSV files of the results and predictions.

Exploratory Data Analysis (EDA)

Enable detailed analysis of the uploaded dataset using Sweetviz. Users can visualize patterns, check correlations, and understand their data before building models.

Upload CSV Files

The user can upload two types of CSV files:

• Training Data: For building multiple machine learning models.
• Prediction Data: For applying the best-performing model to new datasets.

Publication

Divya Karade. (2021). Custom ML Module of AIDrugApp for Molecular Identification, Descriptor Calculation, and Building ML/DL QSAR Models. ChemRxiv Preprint.

How to Run the App

1. Clone this repository.
2. Install the required dependencies using pip install -r requirements.txt.
3. Run the app using the command: streamlit run app.py.
4. The app will open in your default browser.

License

AIDrugApp is licensed under the AGPL-3.0 license. See the LICENSE file for more details.

Contributions

Contributions are welcome. Feel free to submit a pull request or open an issue for discussion.

For more details on how to use the Auto-Multi-ML module, refer to the AIDrugApp.