Overview
RXN Reaction Preprocessing is a tool from the IBM RXN for Chemistry ecosystem for preprocessing chemical-reaction data. It provides standardization, filtering, stable train/validation/test splits, data augmentation and tokenization, with an rxn-data-pipeline command-line entry point and Hydra-based structured configuration for building flexible pipelines.
It is released under MIT and distributed via PyPI (rxn-reaction-preprocessing). It is a data-preprocessing tool independent of RXN-Sandbox.
Key Features
- Standardization, filtering and augmentation of chemical-reaction datasets.
- Stable train/validation/test splits and tokenization.
- `rxn-data-pipeline` entry point with Hydra-based structured configuration.
Use Cases
- Preprocess chemical-reaction datasets for machine-learning tasks such as reaction prediction.
- Build reaction-data pipelines combining standardization, splitting, augmentation and tokenization.
How to Use
Install with pip install rxn-reaction-preprocessing[rdkit] and run via rxn-data-pipeline. See https://rxn4chemistry.github.io/rxn-reaction-preprocessing/.
Sources and verification
This entry describes capabilities documented by the linked project sources. Catalog edits are not evidence of a successful run, compatibility test or scientific performance evaluation.
Read the editorial and verification policy