Towards Docking-oriented De Novo Ligand Design via Gradient Inversion

Zekai Chen, Xunkai Li, Sirui Zhang, Henan Sun, Jia Li, Qiangqiang Dai, Hongchao Qin, Zhenjun Li, Bing Zhou, Rong-Hua Li, Guoren Wang
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:15290-15342, 2026.

Abstract

De novo ligand design is a fundamental task that seeks to generate protein or molecule candidates that can effectively dock with protein receptors and achieve strong binding affinity entirely from scratch. It holds paramount significance for a wide spectrum of biomedical applications. However, most existing studies are constrained by the Pseudo De Novo, Limited Docking Modeling, and Inflexible Ligand Type. To address these issues, we propose MagicDock, a forward-looking framework grounded in the progressive pipeline and differentiable surface modeling. (1) We adopt a well-designed gradient inversion framework. To begin with, general docking knowledge of receptors and ligands is incorporated into the backbone model. Subsequently, the docking knowledge is instantiated as reverse gradient flows by binding prediction, which iteratively guide the de novo generation of ligands. (2) We emphasize differentiable surface modeling in the generation process, leveraging learnable 3D point-cloud representations to precisely capture docking details, thereby ensuring that the generated ligands preserve docking validity through interpretable spatial fingerprints. (3) We introduce customized designs for different ligand types and integrate them into a unified gradient inversion framework with flexible triggers, thereby ensuring broad applicability. Moreover, we provide sufficient theoretical guarantees for MagicDock. Extensive experiments across 9 scenarios demonstrate that MagicDock achieves average improvements of 7.0% and 7.4% over SOTA baselines specialized for protein or molecule ligand design, respectively.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-chen26bu, title = {Towards Docking-oriented De Novo Ligand Design via Gradient Inversion}, author = {Chen, Zekai and Li, Xunkai and Zhang, Sirui and Sun, Henan and Li, Jia and Dai, Qiangqiang and Qin, Hongchao and Li, Zhenjun and Zhou, Bing and Li, Rong-Hua and Wang, Guoren}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {15290--15342}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/chen26bu/chen26bu.pdf}, url = {https://proceedings.mlr.press/v306/chen26bu.html}, abstract = {De novo ligand design is a fundamental task that seeks to generate protein or molecule candidates that can effectively dock with protein receptors and achieve strong binding affinity entirely from scratch. It holds paramount significance for a wide spectrum of biomedical applications. However, most existing studies are constrained by the Pseudo De Novo, Limited Docking Modeling, and Inflexible Ligand Type. To address these issues, we propose MagicDock, a forward-looking framework grounded in the progressive pipeline and differentiable surface modeling. (1) We adopt a well-designed gradient inversion framework. To begin with, general docking knowledge of receptors and ligands is incorporated into the backbone model. Subsequently, the docking knowledge is instantiated as reverse gradient flows by binding prediction, which iteratively guide the de novo generation of ligands. (2) We emphasize differentiable surface modeling in the generation process, leveraging learnable 3D point-cloud representations to precisely capture docking details, thereby ensuring that the generated ligands preserve docking validity through interpretable spatial fingerprints. (3) We introduce customized designs for different ligand types and integrate them into a unified gradient inversion framework with flexible triggers, thereby ensuring broad applicability. Moreover, we provide sufficient theoretical guarantees for MagicDock. Extensive experiments across 9 scenarios demonstrate that MagicDock achieves average improvements of 7.0% and 7.4% over SOTA baselines specialized for protein or molecule ligand design, respectively.} }
Endnote
%0 Conference Paper %T Towards Docking-oriented De Novo Ligand Design via Gradient Inversion %A Zekai Chen %A Xunkai Li %A Sirui Zhang %A Henan Sun %A Jia Li %A Qiangqiang Dai %A Hongchao Qin %A Zhenjun Li %A Bing Zhou %A Rong-Hua Li %A Guoren Wang %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-chen26bu %I PMLR %P 15290--15342 %U https://proceedings.mlr.press/v306/chen26bu.html %V 306 %X De novo ligand design is a fundamental task that seeks to generate protein or molecule candidates that can effectively dock with protein receptors and achieve strong binding affinity entirely from scratch. It holds paramount significance for a wide spectrum of biomedical applications. However, most existing studies are constrained by the Pseudo De Novo, Limited Docking Modeling, and Inflexible Ligand Type. To address these issues, we propose MagicDock, a forward-looking framework grounded in the progressive pipeline and differentiable surface modeling. (1) We adopt a well-designed gradient inversion framework. To begin with, general docking knowledge of receptors and ligands is incorporated into the backbone model. Subsequently, the docking knowledge is instantiated as reverse gradient flows by binding prediction, which iteratively guide the de novo generation of ligands. (2) We emphasize differentiable surface modeling in the generation process, leveraging learnable 3D point-cloud representations to precisely capture docking details, thereby ensuring that the generated ligands preserve docking validity through interpretable spatial fingerprints. (3) We introduce customized designs for different ligand types and integrate them into a unified gradient inversion framework with flexible triggers, thereby ensuring broad applicability. Moreover, we provide sufficient theoretical guarantees for MagicDock. Extensive experiments across 9 scenarios demonstrate that MagicDock achieves average improvements of 7.0% and 7.4% over SOTA baselines specialized for protein or molecule ligand design, respectively.
APA
Chen, Z., Li, X., Zhang, S., Sun, H., Li, J., Dai, Q., Qin, H., Li, Z., Zhou, B., Li, R. & Wang, G.. (2026). Towards Docking-oriented De Novo Ligand Design via Gradient Inversion. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:15290-15342 Available from https://proceedings.mlr.press/v306/chen26bu.html.

Related Material