In AutoDock software, you need to perform docking one ligand at a time. If you want to dock multiple ligands simultaneously, you can use PyRx software, which allows batch docking of multiple ligands and is more user-friendly.
AutoDock and AutoDock Vina can effectively handle multiple ligands in docking simulations through sequential processing, though true simultaneous multi-ligand docking requires specific adaptations. The standard approach involves preparing individual ligand files in .pdbqt format using tools like Open Babel to ensure proper protonation states and charges, followed by running separate docking jobs for each compound with defined grid box parameters centered on the target binding site. For large ligand libraries, batch processing can be automated through Python or Bash scripts to sequentially submit jobs. While AutoDock itself doesn't natively support co-docking of multiple ligands in a single simulation, users can manually merge ligands into a single .pdbqt file and expand the search space, though this may compromise scoring accuracy since the force field isn't optimized for ligand-ligand interactions. More robust solutions for studying cooperative binding involve combining initial AutoDock results with molecular dynamics simulations in packages like GROMACS or AMBER to refine poses and account for dynamic effects. Specialized platforms like HADDOCK explicitly support multi-component docking through flexible protocols that accommodate protein-ligand-ligand complexes. Critical considerations include careful preparation of ligand files to ensure correct atom typing and charge assignment, appropriate grid box sizing to encompass potential binding regions, and validation of results through experimental data or complementary computational methods like MM-PBSA calculations. The workflow typically begins with ligand preparation through structure conversion and optimization, followed by systematic docking runs either locally or on high-performance computing clusters, with post-processing analysis to identify promising candidates for further investigation. While AutoDock provides efficient screening capabilities, researchers studying complex multi-ligand binding scenarios may need to integrate additional modeling techniques to achieve biologically relevant results. Commercial alternatives like Schrödinger's Glide or MOE offer more sophisticated multi-ligand handling but require licensing, making AutoDock/Vina combined with custom scripting a practical open-source solution for most virtual screening needs.