STOUT V2.0: SMILES to IUPAC Name Conversion

Paper Information

Citation: Rajan, K., Zielesny, A., & Steinbeck, C. (2024). STOUT V2.0: SMILES to IUPAC name conversion using transformer models. Journal of Cheminformatics, 16(146). https://doi.org/10.1186/s13321-024-00941-x

Publication: Journal of Cheminformatics 2024

Additional Resources:

• Web Application
• DECIMER Project

What kind of paper is this?

Method (Primary) / Resource (Secondary)

This paper presents a Methodological contribution by developing and validating a Transformer-based neural machine translation model (STOUT V2) for chemical nomenclature. It systematically compares this new architecture against previous RNN-based baselines (STOUT V1) and performs ablation studies on tokenization strategies.

It also serves as a significant Resource contribution by generating a massive training dataset of nearly 1 billion SMILES-IUPAC pairs (curated via commercial Lexichem software) and releasing the resulting models and code as open-source tools to democratize access to high-quality chemical naming.

What is the motivation?

Assigning systematic IUPAC names to chemical structures is governed by complex, comprehensive rules that are difficult for humans to apply consistently. While deterministic, rule-based software (e.g., OpenEye Lexichem, ChemAxon) exists and is reliable, it is often commercial and not freely available to all researchers. Existing open-source tools like OPSIN are primarily for parsing names to structures, not generating names.

The previous version of STOUT (V1), based on RNNs/GRUs, showed promise but had limitations in accuracy (BLEU ~90%) and handling stereochemistry. This work aims to leverage the superior sequence-learning capabilities of Transformers and massive datasets to create a highly accurate, open-source IUPAC naming tool.

What is the novelty here?

The core novelty lies in the architectural shift and the unprecedented scale of training data:

1. Architecture Shift: Moving from an RNN-based Seq2Seq model to a Transformer-based architecture (4 layers, 8 heads), which captures intricate chemical patterns better than GRUs.
2. Billion-Scale Training: Training on a dataset of nearly 1 billion molecules (combining PubChem and ZINC15), significantly larger than the 60 million used for STOUT V1.
3. Tokenization Strategy: Determining that character-wise tokenization for IUPAC names is superior to word-wise tokenization in terms of both accuracy and training efficiency (15% faster).

What experiments were performed?

The authors conducted three primary experiments to validate the model:

• Experiment 1 (Optimization): Assessed the impact of dataset size (1M vs 10M vs 50M) and tokenization strategy (word-split vs character-split) on performance.
• Experiment 2 (Scaling): Trained a model on 110 million PubChem molecules to test performance on longer sequences (up to 600 chars).
• Experiment 3 (Generalization): Trained on the full ~1 billion dataset (PubChem + ZINC15).
• External Validation: Benchmarked against an external dataset from ChEBI (1,485 molecules) and ChEMBL34 to test generalization to unseen data and long chemical chains.

Evaluation Metrics:

• Textual Accuracy: BLEU scores (1-4) and Exact String Match.
• Chemical Validity: Retranslation of generated names back to SMILES using OPSIN, followed by Tanimoto similarity checks (PubChem fingerprints) against the original input.

What were the outcomes and conclusions drawn?

• Superior Accuracy: STOUT V2 achieved an average BLEU score of 0.99 (vs 0.94 for V1) and perfect exact matches on 97.49% of the test set (vs 66.65% for V1).
• Tokenization: Character-level splitting for IUPAC names outperformed word-level splitting and was more computationally efficient.
• Generalization: The 1-billion-scale model significantly outperformed the 100-million-scale model on longer, more complex sequences (SMILES > 300 chars), maintaining high Tanimoto similarity where the smaller model failed.
• Limitations: Performance drops for very long SMILES (>600 characters) due to their scarcity in training data. The model depends on the quality of the commercial software (Lexichem) used to generate ground truth.

Reproducibility Details

Data

The training data was derived from PubChem and ZINC15. Ground truth IUPAC names were generated using OpenEye Lexichem TK 2.8.1 to ensure consistency.

Preprocessing:

• SMILES: Canonicalized, isomeric, and kekulized using RDKit (v2023.03.1).
• Formatting: Converted to TFRecord format in 100 MB chunks for TPU efficiency.

Algorithms

• SMILES Tokenization: Regex-based splitting. Atoms (e.g., “Cl”, “Au”), bonds, brackets, and digits are separate tokens.
• IUPAC Tokenization: Character-wise split was selected as the optimal strategy (treating every character as a token).
• Optimization: Adam optimizer with a custom learning rate scheduler based on model dimensions.
• Loss Function: Sparse Categorical Cross-Entropy, masking padding tokens.

Models

The model follows the standard Transformer architecture from “Attention is All You Need” (Vaswani et al.).

• Architecture: 4 Transformer layers (encoder/decoder stack).
• Attention: Multi-head attention with 8 heads.
• Dimensions: Embedding size ($d_{model}$) = 512; Feed-forward dimension ($d_{ff}$) = 2048.
• Regularization: Dropout rate of 0.1.
• Context Window: Max input length (SMILES) = 600; Max output length (IUPAC) = 700-1000.

Evaluation

Evaluation focused on both string similarity and chemical structural integrity.

Hardware

Training was conducted entirely on Google Cloud Platform (GCP) TPUs.

• STOUT V1: Trained on TPU v3-8.
• STOUT V2: Trained on TPU v4-128 pod slices (128 nodes).
• Large Scale (Exp 3): Trained on TPU v4-256 pod slice (256 nodes).
• Training Time: Average of 15 hours and 2 minutes per epoch for the 1 billion dataset.
• Framework: TensorFlow 2.15.0-pjrt with Keras.

Citation

@article{rajanSTOUTV20SMILES2024,
  title = {{{STOUT V2}}.0: {{SMILES}} to {{IUPAC}} Name Conversion Using Transformer Models},
  shorttitle = {{{STOUT V2}}.0},
  author = {Rajan, Kohulan and Zielesny, Achim and Steinbeck, Christoph},
  year = 2024,
  month = dec,
  journal = {Journal of Cheminformatics},
  volume = {16},
  number = {1},
  pages = {146},
  issn = {1758-2946},
  doi = {10.1186/s13321-024-00941-x}
}