SimNMR-PubChem Database

🧬 Overview

The SimNMR-PubChem Database provides a comprehensive collection of 106 million molecules from PubChem, each paired with corresponding simulated NMR spectra. The dataset has been carefully preprocessed and optimized for NMR spectral analysis, molecular structure elucidation, and machine learning applications in computational chemistry.

📁 Dataset Structure

The database is organized into two main components:

Metadata Storage (metadata/)

• Format: LMDB (Lightning Memory-Mapped Database)
• Content: Complete molecular metadata including:
  • SMILES representations
  • Simulated ¹H and ¹³C NMR spectra
  • Molecular formulas and InChIKeys
  • 3D conformers
  • Additional molecular and spectral properties

Vector Index (index/)

• Format: Faiss index file
• Content: High-dimensional vector embeddings of NMR spectra
• Purpose: Enables ultra-fast similarity search and nearest neighbor retrieval

🚀 Quick Start

Step 1: Merge Database Parts

Due to file size constraints, the dataset is split into multiple parts. Merge them using the following commands:

# Merge LMDB database
cat metadata/PubChem_merged_id.lmdb.part-* > metadata/PubChem_merged_id.lmdb

# Merge Faiss index
cat index/HC.index.part-* > index/HC.index

Step 2: Usage Examples

For detailed usage and examples, please refer to the GitHub repository where the full code has been open-sourced. You can find scripts to deploy and query the database, as well as how to perform similarity searches using the Faiss index. The repository incorporates the database into the NMR-Solver project, a framework for automated molecular structure elucidation.

⚠️ Requirements

• Storage: Ensure sufficient disk space (~500 GB total)
• Memory: Recommended 128GB+ RAM for optimal performance

🔗 Resources

• Source Code: NMR-Solver GitHub Repository
• Research Paper: arXiv:2509.00640

📖 Citation

If you use this database in your research, please kindly cite our work:

@article{jin2025nmrsolver,
  title={NMR-Solver: Automated Structure Elucidation via Large-Scale Spectral Matching and Physics-Guided Fragment Optimization},
  author={Jin, Yongqi and Wang, Jun-Jie and Xu, Fanjie and Ji, Xiaohong and Gao, Zhifeng and Zhang, Linfeng and Ke, Guolin and Zhu, Rong and E, Weinan},
  year={2025},
  journal={arXiv preprint arXiv:2509.00640}
}