Steering Semi-Flexible Molecular Diffusion Model for Structure-Based Drug Design

SeFMol is an online platform for protein-pocket based 3D molecule generation. You can submit a target pocket, set property preferences, and quickly obtain candidate molecules with downloadable results for follow-up screening.

Step 1 Upload protein

Step 2 Set constraints

Step 3 Run generation

Step 4 Pick candidates

Get Started >>

Functional Highlights

Protein-pocket conditioned 3D molecular generation.
RL-steered semi-flexible conformational adjustment.
Multi-property guidance for controllable sampling.
Fast sampling: reduced from 1000 to 50 diffusion steps.
End-to-end web workflow for upload, generation, visualization, and download.

Please Cite

Xudong Zhang, Sanqing Qu, Fan Lu, Jianmin Wang, Zhixin Tian, Shangding Gu, Yanping Zhang, Alois Knoll, Shaorong Gao, Guang Chen, Changjun Jiang, Steering Semi-Flexible Molecular Diffusion Model for Structure-Based Drug Design with Reinforcement Learning, Science Advances, 2026.

Xudong Zhang, Jing Hou, Sanqing Qu, Fan Lu, Zhixin Tian, Alois Knoll, Guang Chen, Shaorong Gao, Yanping Zhang, Deep reinforcement learning as an interaction agent to steer fragment-based 3D molecular generation for protein pockets, Briefings in Bioinformatics, 2025.