DockIt is fast. The default settings represent a compromise between speed and complete sampling. Nevertheless, we believe that DockIt is faster than other flexible ligand docking programs (it takes between 30 seconds and one minute to generate 100 docked conformations of a typical ligand) By modification of the parameters, it is possible to have DockIt run considerably faster. We have also spent considerable effort in making the program easy to set up and use. In particular, the initial set up for a new protein target is generally automatic and only takes a few minutes. Of course, like all programs, there are times when it is necessary to tune the parameters to DockIt to get the best results. However, we have tried to build intelligence into DockIt to get good results as automatically as possible.

A crucial part of current docking programs is the scoring function used to evaluate ligand-receptor complementarity. We believe that the scoring function determines, to a large extent, the quality of the results. DockIt comes with three scoring functions: a molecular mechanics function (based roughly on Amber non-bonded terms), Potential of Mean Force (PMF, from the Abbott group) and Piecewise Linear Potential (PLP from the Agouron group). These can be used in combination in various consensus schemes which seem to give better results than any single function. In addition, DockIt1.5, includes the capability for the user to easily add other scoring functions and select them and/or the built in functions at run time. In addition to providing the ability to add additional general purpose scoring functions, this will allow for functions to be generated for special cases (for example a particular family of proteins).

DockIt is well adapted to parallel computations. It comes with a PVM version which works very well on loosely coupled clusters (like the Linux clusters which are becoming very common). The parallel version of the program is run in exactly the same way as the single processor version (single input stream and single output stream) so it is quite easy to do large runs.

Finally, note that DockIt uses a distance geometry based approach to docking. DockIt only requires 2D connectivity information about the ligands (not 3D structures) so it is quite easy to output some connection table type information (Daylight SMILES or MDL mol files) from a database or a combinatorial design computation through DockIt. It is not necessary to compute 3D coordinates of the ligand in order to use DockIt. In addition, this means that rings are flexible (not just torsions as in most docking programs) and if stereochemistry is unknown for any centers, the possible stereochemistries are randomly sampled. Of course, DockIt respects any stereochemical information which is provided. The distance geometry engine in DockIt first generates a bound conformation of the ligand which fits into the shape of the binding pocket and then refines both shape and chemical complementarity. This is different from many other docking programs which first get a few electrostatic interactions (hydrogen bonds typically) placed and then refine the shape complementarity. Thus, DockIt is complementary to most other docking programs.