MolGlyph

This is the official implement of MolGlyph, an open-source optical chemical structure reconigtion (OCSR) model developed within the BioMiner project

OCSR Performance on BioVista:

Structure Types	MolMiner	MolScribe	MolNexTR	DECIMER	MolParser	MolGlyph (Ours)
Full	0.774	0.703	0.695	0.545	0.669	0.758
Chiral	0.497	0.481	0.419	0.326	0.352	0.504
Markush	0.185	0.156	0.045	0.000	0.733	0.770
All	0.507	0.455	0.401	0.298	0.703	0.764

Usage

Download the checkpoint from this repo and see details in BioMiner for usage.

Contact

We welcome all forms of feedback! Please raise an issue for bugs, questions, or suggestions. This helps our team address common problems efficiently and builds a more productive community. If you encounter any issues, please reach out to [email protected].

Citation

If you find our work helpful, please kindly cite:

@article {Yan2025.04.22.648951,
    author = {Yan, Jiaxian and Zhu, Jintao and Yang, Yuhang and Liu, Qi and Zhang, Kai and Zhang, Zaixi and Liu, Xukai and Zhang, Boyan and Gao, Kaiyuan and Xiao, Jinchuan and Chen, Enhong},
    title = {BioMiner: A Multi-modal System for Automated Mining of Protein-Ligand Bioactivity Data from Literature},
    year = {2025},
    doi = {10.1101/2025.04.22.648951},
    journal = {bioRxiv}
}