1. Introduction

The first version of AutoDock 1 was distributed to over 35 sites around the world, and that number has since grown to over 600 sites with the latest versions of AutoDock 2 ,3. This user guide is the first version to accompany a significantly enhanced version of AutoDock, version 3.0, which includes powerful new search methods and a new empirical free energy function 3 .

The program AutoDock was developed to provide an automated procedure for predicting the interaction of ligands with biomacromolecular targets. The motivation for this work arises from problems in the design of bioactive compounds, and in particular the field of computer-aided drug design. Progress in biomolecular x-ray crystallography continues to provide a number of important protein and nucleic acid structures. These structures could be targets for bioactive agents in the control of animal and plant diseases, or simply key to understanding of a fundamental aspect of biology. The precise interaction of such agents or candidate molecules is important in the development process. Indeed, AutoDock can be a valuable tool in the x-ray structure determination process itself: given the electron density for a ligand, AutoDock can help to narrow the conformational possibilities and help identify a good structure. Our goal has been to provide a computational tool to assist researchers in the determination of biomolecular complexes.

In any docking scheme two conflicting requirements must be balanced: the desire for a robust and accurate procedure, and the desire to keep the computational demands at a reasonable level. The ideal procedure would find the global minimum in the interaction energy between the substrate and the target protein, exploring all available degrees of freedom (DOF) for the system. However, it must also run on a laboratory workstation within an amount of time comparable to other computations that a structural researcher may undertake, such as a crystallographic refinement. In order to meet these demands a number of docking techniques simplify the docking procedure. Still one of the most common techniques in use today is manually-assisted docking. Here, the internal and orientational degrees of freedom in the substrate are under interactive control. While the energy evaluation for such techniques can be sophisticated, the global exploration of configurational space is limited. At the other end of the spectrum are automated methods such as exhaustive search and distance geometry. These methods can explore configurational space, but at the cost of a much simplified model for the energetic evaluation.

The original procedure developed for AutoDock used a Monte Carlo (MC) simulated annealing (SA) technique for configurational exploration with a rapid energy evaluation using grid-based molecular affinity potentials. It thus combined the advantages of exploring a large search space and a robust energy evaluation. This has proven to be a powerful approach to the problem of docking a flexible substrate into the binding site of a static protein. Input to the procedure is minimal. The researcher specifies a rectangular volume around the protein, the rotatable bonds for the substrate, and an arbitrary or random starting configuration, and the procedure produces a relatively unbiased docking.