Water Solubility Prediction

A web based application predicts water solubility of any given chemical compound whether known or unknown.

Introduction • Requirements • Installation • Usage • How it Works • Thanks ❤

Introduction

This Project is inspired from a research paper by Delaney published in 2003, Predicting water solubility of any chemical compound. Here we train our model on four features suggested by delaney.

Requirements

Python 3.3+
macOs or Linux or Windows

Installation

Building the source code

1. Clone the repository

git clone https://github.com/arhamshah/SolubilityPrediction.git
cd SolubilityPrediction

2. Create Anaconda Environment

First we create an environment

conda create -n ml-project python=3.8

We need to activate that environment

conda activate ml-project

3. Download & Install all the Dependencies

pip install -r requirements.txt

For installation of rdkit refer here.

Usage

Checkout Video Tutorial

Input SMILES

Enter SMILES code of compound. For Multiple Entries, each compound on a new line.
Press Ctrl + Enter to calculate.

How it Works

Checkout research paper by Delaney here

Descriptors such as Molecular LogP, Molecular Weight, Number of Rotatable Bonds, Aromatic Proportion are calculated.
Model is trained using CatBoost regressor.
Solubility prediction is made with input compound given by user.
Streamlit is used for Graphical User Interface and Deployment.
Calculated accuracy is about 93.35%.

Thanks

Delaney for providing insight from research-paper.
Data Professor for helping me build this project.
Shoutout to developers & contributors of Pillow, Pip, Pycaret, Rdkit.

sailfish009/SolubilityPrediction