visChem 0.1

A tool for chemical space visualization and clustering.

visChem v0.1

Steve Niu[^1]

visChem is a Python module designed for visualizing chemical space using UMAP and clustering techniques. This tool facilitates the transformation of chemical structures into a 2D space and visualizes clusters of similar compounds.

Installation

1. Clone the repository:

git clone visChem  | cd visChem

2. Install the required packages:

pip install -r requirements.txt

Note: Some packages like rdkit might require special installation instructions. Please refer to their official documentation for details.

Usage

1. Prepare your data file, e.g., 'NCI/first_200.props.sdf'.
2. Run the example.py script:

python example.py

This script will execute the main function which:

• Reads the SMILES from the data file.
• Converts SMILES to molecular representations.
• Generates molecular fingerprints.
• Computes Tanimoto similarity.
• Reduces dimensionality using UMAP.
• Performs clustering.
• Visualizes the chemical structures in the clustered chemical space below.

Refer to example.py for details and any potential modifications you might want to make for your specific use case.

[^1]: Safety Assessment, [e-mail]