How do I use Molecular Transformer to predict reactions for my own molecules?

Preprocess your SMILES data with the provided tokenizer, format it to match the dataset structure, and run the translate.py script with a trained model. For custom reactions, you may need to retrain or fine-tune using the training commands, which requires adapting the data pipeline.

Molecular Transformer vs IBM RXN: which is better for everyday use?

Molecular Transformer is open-source and customizable for research with public data, while IBM RXN offers a user-friendly GUI and models trained on more diverse, possibly proprietary data. Choose based on whether you need flexibility or ease of use without coding.

What hardware is needed to train Molecular Transformer from scratch?

The README specifies training on a single GPU for 48-72 hours, so a modern GPU with sufficient memory is essential. Inference can run on CPUs but will be slower, and specific CUDA versions are required for the outdated PyTorch setup.

Can I fine-tune the pre-trained models on my proprietary dataset?

Yes, but it requires adapting the data preprocessing scripts to your format and retraining with the provided commands. This process is complex and assumes familiarity with the codebase and machine learning workflows.

Is there a web API or demo for Molecular Transformer?

No, the repository only provides code for local deployment; for a web interface, use IBM RXN. Deploying as a service would require building a custom API wrapper, which isn't included in the project.

Molecular Transformer — Chemical Reaction Predictor

What is Molecular Transformer?

Molecular Transformer is a sequence-to-sequence neural network model that predicts chemical reaction outcomes and retrosynthetic pathways. It treats molecules as SMILES strings and uses transformer architecture to translate between reactants and products, helping chemists design synthesis routes faster. The model includes uncertainty estimation to indicate prediction confidence.

Target Audience

Computational chemists, researchers in cheminformatics, and organic chemists who need AI tools for reaction prediction and retrosynthesis planning.

Value Proposition

It provides an open-source, uncertainty-calibrated model trained on public reaction datasets, unlike proprietary tools. The integration with RDKit for data preprocessing and availability of pre-trained models lowers the barrier for academic and industrial adoption.

Overview

Molecular Transformer is a neural machine translation model adapted for chemistry that predicts chemical reaction outcomes and retrosynthetic pathways. It translates between molecular representations (SMILES strings) to forecast how molecules react or how target molecules can be synthesized, accelerating discovery in organic chemistry and drug development.

Key Features

Retrosynthesis Prediction — Predicts reactant molecules needed to synthesize a target product molecule.
Uncertainty Calibration — Provides confidence estimates for predictions, helping chemists assess reliability.
SMILES Tokenization — Uses custom tokenization of SMILES strings to treat molecules as sequences for transformer models.
Data Augmentation — Doubles training data by generating random equivalent SMILES representations via RDKit.
Pre-trained Models — Includes models trained on public datasets (USPTO_MIT, USPTO_STEREO) with mixed or separated reactant/reagent formats.

Philosophy

Molecular Transformer aims to make AI-assisted chemical reaction prediction accessible to organic chemists, with the goal of integrating these models into daily laboratory workflows to accelerate molecular discovery.

Molecular Transformer

What is Molecular Transformer?

Overview

Key Features

Philosophy

Related Projects

Found a gem we're missing?

Use Cases

Best For

Not Ideal For

Pros & Cons

Pros

Cons

Frequently Asked Questions