Generative Virtual Screening Toolkit

This repository introduces a proof-of-concept toolkit for generative virtual screening, combining seed-based molecule generation, QSAR predictions, retrosynthesis, and reaction yield estimation powered by Hugging Face transformer models.

The repository contains the following Python modules and notebooks:

Repository Structure

• ChemBERT_module.py
  An OOP-based module for molecule generation using the ChemBERTaLM model
  (https://huggingface.co/gokceuludogan/ChemBERTaLM).

• QSAR.ipynb
  A Jupyter notebook demonstrating basic data preprocessing and AutoML-based QSAR model construction.

• ReactionT5Retrosynthesis_module.py
  An OOP-based module for retrosynthesis prediction using the
  ReactionT5v2-retrosynthesis-USPTO_50k model
  (https://huggingface.co/sagawa/ReactionT5v2-retrosynthesis-USPTO_50k).

• ReactionT5Yield.py
  An OOP-based module for reaction yield prediction using the
  ReactionT5v2-yield model
  (https://huggingface.co/sagawa/ReactionT5v2-yield).

• VS_tool.ipynb
  The main notebook implementing the generative virtual screening workflow, integrating all functionalities provided in this repository.

• environment.yml YAML file with exported environment.

• output_visualization.py
  A lightweight and elegant script for visualizing the generated results.

---

If you find this repository interesting or useful, feel free to ⭐ star it!