ColabDock is a deep-learning protein–protein docking framework designed to predict how two proteins interact and to identify plausible three-dimensional complex structures. Developed by adapting deep-learning structure-prediction approaches for the docking problem, ColabDock combines generation of alternative protein-complex conformations with structure prediction and model ranking. This approach allows the docking process to explore multiple possible interaction arrangements rather than relying solely on conventional rigid-body sampling. By leveraging the structural prediction capabilities of AlphaFold2 within a dedicated docking workflow, ColabDock provides a modern computational approach for investigating protein–protein interactions and generating structural models of protein complexes.

• Deep-learning docking: Predicts three-dimensional protein–protein complex structures.
• Alternative complex generation: Explores multiple possible interaction arrangements.
• AlphaFold2-based prediction: Uses AlphaFold2 within a dedicated protein–protein docking workflow.
• Model ranking: Evaluates candidate complexes to identify promising docking solutions.
• Protein–protein interaction modeling: Designed specifically for predicting protein complexes, including antibody–antigen interactions.

Created by

Shihao Feng, Zhenyu Chen, Chengwei Zhang, Yuhao Xie, Sergey Ovchinnikov, Yi Qin Gao, and Sirui Liu.

Selected research publications relevant to ColabDock.