Adhering to the ligand-pharmacophore mapping principles, we proposed DiffPhore, a pioneering knowledge-guided diffusion framework for 'on-the-fly' 3D ligand-pharmacophore mapping. The main concept behind DiffPhore is to utilize ligand-pharmacophore mapping principles to guide the generative process, meanwhile leveraging calibrated sampling to mitigate the exposure bias inherent in diffusion models. The source code, instructions, model weights and datasets for training (LigPhoreSet, CpxPhoreSet) can be accessed via here.
Here's a video demo of 3D-pharmacophore mapping driven by DiffPhore:
We sincerely are open to receiving support and advice from academic and industrial communities to improve DiffPhore's usefulness, please email us: ddtmlab_gbl@sina.com.
Yu, J.; Zhou, C.; Ning, X.; Mou, J.; Meng, F.; Wu, J.; Chen, Y.; Liu, X.*; Li, G.-B*. Knowledge-Guided Diffusion Model for 3D Ligand-Pharmacophore Mapping (in preparation)
*To whom correspondence should be addressed.