Skip to content

About

Physics-based machine learning approach to predict hydration free energies for small molecules, combining data-driven methods with physics-informed insights for accurate and reliable predictions. Includes EDA, data grouping, and advanced regression modeling tools.

Resources

Stars

1 star

Watchers

1 watching

Forks

Repository files navigation

Physics-based Machine Learning to Predict Hydration Free Energies for Small Molecules

Overview

This research project explores the application of physics-informed machine learning techniques to accurately predict hydration free energies for small molecules. The methodology combines data-driven approaches with physics-based insights to enhance predictive capabilities.

Notebooks Included

  1. eda_pp.ipynb: Conducts exploratory data analysis (EDA) and preprocessing, using visualization tools like matplotlib and seaborn to understand molecular datasets.
  2. make_groups.ipynb: Organizes molecules into meaningful categories, leveraging structural or chemical properties to aid in model training.
  3. regression.ipynb: Implements and evaluates regression models tailored to predict hydration free energies, using metrics like mean absolute error (MAE), mean squared error (MSE), and R-squared.

Features

  • Physics-informed EDA: Utilizes domain knowledge for effective data preparation and analysis.
  • Data Structuring: Tools for categorizing molecules based on features critical to hydration energy predictions.
  • Advanced Regression Models: Machine learning models that integrate physics-based principles to improve accuracy and reliability.

Prerequisites

Ensure the following are installed. It is highly recommended to use the provided environment.yml file to create a consistent environment.

  • Python 3.12.2
  • The following libraries, with the specified versions:
    • numpy==1.26.4
    • scipy==1.15.0
    • pandas==2.2.3
    • scikit-learn==1.6.1
    • xgboost==2.1.3
    • lightgbm==4.5.0
    • plotly==5.24.1
    • matplotlib==3.10.0
    • fastapi==0.115.7

Environment Setup (Recommended)

Miniforge3)

  1. Using conda:

    Conda installation using miniforge -

    curl -L -O "https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-$(uname)-$(uname -m).sh"
    
    bash Miniforge3-$(uname)-$(uname -m).sh

    Windows - Download Miniforge3 for Windows (64-bit)

    It is highly recommended to use conda to create an environment with the specified dependencies. This ensures that you have the correct versions of all the required packages.

    • Create the environment from the environment.yml file:

      conda env create -f environment.yml
    • Activate the environment:

      conda activate scis
  2. Alternative: Using pip (Not Recommended):

    While not recommended, you can attempt to install the dependencies using pip. However, this may lead to version conflicts or other issues.

    • Create a requirements.txt file with the following content:

      numpy==1.26.4
      scipy==1.15.0
      pandas==2.2.3
      scikit-learn==1.6.1
      xgboost==2.1.3
      lightgbm==4.5.0
      plotly==5.24.1
      matplotlib==3.10.0
      fastapi==0.115.7
    • Install the dependencies:

      pip install -r requirements.txt

Usage

  1. EDA and Preprocessing:

    Open eda_pp.ipynb to:

    • Visualize molecular data distributions.
    • Perform feature engineering and data preprocessing.
  2. Data Grouping:

    Use make_groups.ipynb for:

    • Organizing molecules based on their physical or chemical properties.
    • Structuring data to enhance model training and evaluation.
  3. Regression Modeling:

    Run regression.ipynb to:

    • Train machine learning models on hydration free energy datasets.
    • Evaluate model performance using key metrics.

Contributions

Contributions to this research are welcome. Suggestions, code improvements, and additional experiments can be submitted via pull requests.

Authors

Marvin V Prakash, Ajeet Yadav, Dr. Pradipta Bandyopadhyay

Reference Links

Add the relevant links here, such as:

About

Physics-based machine learning approach to predict hydration free energies for small molecules, combining data-driven methods with physics-informed insights for accurate and reliable predictions. Includes EDA, data grouping, and advanced regression modeling tools.

Resources

Stars

1 star

Watchers

1 watching

Forks

Releases

Packages

Contributors

Languages