smilesfeaturizer 0.1.2

A Python package that automatically generates derived variables from a column with SMILES (Simplified Molecular-Input Line-Entry System).

Development Status :: 3 - Alpha

SMILES featurizer

A Python package that automatically generates derived variables from a column with SMILES (Simplified Molecular-Input Line-Entry System)

Install

$ pip install smilesfeaturizer

$ pip install git+https://github.com/dsdanielpark/SMILES-featurizer.git

Usage

The dataset assumes the presence of SMILES strings in a column named SMILES.

Feature generation

• Create fingerprint columns for SMILES representations based on various packages RDKit, Mol2Vec, DataMol, MolFeat, Scikit-Learn.

  from smilesfeaturizer import generate_smiles_feature
  
  df = generate_smiles_feature(df) # default method="simple"
  
  df = generate_smiles_feature(df, method="specific") 
  

Create dashboard

• Through the dashboard, you can determine which compounds exhibit what prediction performance.

  from smilesfeaturizer import create_inline_dash_dashboard
  
  # Load your DataFrame and specify the columns
  true_col = 'pIC50'
  predicted_col = 'predicted_pIC50'
  
  # Create and run the Dash dashboard
  create_inline_dash_dashboard(df, true_col, predicted_col)
  

Save reporting images

• Molecular images, basic information, and the prediction versus actual values are visually represented in bar graphs for easy viewing.

  from smilesfeaturizer import smiles_insight_plot
  
  selected_metric = 'RMSE'  # Choose the error metric you want to display
  true_col = 'pIC50'  # Replace with your true column name
  predicted_col = 'predicted_pIC50'  # Replace with your predicted column name
  smiles_insight_plot(df[:1], true_col, predicted_col, selected_metric, 'output_folder', show=True)
  

License

Apache 2.0

Bugs and Issues

Sincerely grateful for any reports on new features or bugs. Your valuable feedback on the code is highly appreciated.

Contacts

• Core maintainer: Daniel Park, South Korea
  
• E-mail: [e-mail]
  

Copyright (c) 2023 MinWoo Park, South Korea