Skip to content
Leonardo Solis V edited this page Dec 12, 2021 · 24 revisions


Welcome to the AutoDock-GPU wiki!

AutoDock-GPU is an OpenCL implementation of AutoDock LGA.

Main project is hosted under:


AutoDock-GPU leverages the embarrassingly parallelism of AutoDock LGA, by processing ligand-receptor poses in parallel over multiple compute units on a GPU device.

Besides the legacy Solis-Wets local search method, AutoDock-GPU adds newly implemented local-search methods based on gradients of the scoring function.

One of these methods, ADADELTA, has proven to increase significantly the docking quality in terms of scores and RMSDs.


AutoDock-GPU is released under the GNU GPL v2 and GNU LGPL v2.1 licenses. See details.



A paper describing the OpenCL implementation and evaluation of AutoDock-GPU is available on the Journal of Chemical Theory and Computation

If you find AutoDock-GPU useful for your work, please cite it as:

  title = {Accelerating AutoDock4 with GPUs and Gradient-Based Local Search},
  author = {Santos-Martins, Diogo and Solis-Vasquez, Leonardo and Tillack, Andreas F and Sanner, Michel F and Koch, Andreas and Forli, Stefano},
  journal = {Journal of Chemical Theory and Computation},
  volume = {17},
  number = {2},
  pages = {1060-1073},
  year = {2021},
  doi = {10.1021/acs.jctc.0c01006},
  publisher = {ACS Publications}

The preprint preceeding this publication is available on ChemRxiv using an earlier version (v1.1) of AutoDock-GPU.

A full list of publications is available here.


If you encounter any issues or have any questions, please use any of the following:

You can also contact the main developers via email: