Two modes of energy calculation are available: continuum or grid-based.Continuum energies are the more accurate of the two, as they are calculated on-the-fly based on exact distance calculations to nearby atoms. Grid-based energies are less accurate as they are computed by simple summation of terms from precalculated grids, most commonly employing linear interpolation. The energies obtained via both methods are consistent with each other. As one might expect, grid-based computation is very rapid, typically 50 times faster than the analogous continuum calculation.
Two different minimization techniques are also available: quasi-Newton (dockmin_dfp) and simplex (dockmin_sim). The quasi-Newton minimizer is based on that written by Jeff Blaney (1982) and uses numerical derivatives within a modified Davidon-Fletcher-Powell routine (Fletcher, 1960). The simplex minimizer is that of Nelder and Mead (1965). On grid calculations, simplex is at least twice as fast: quasi-Newton optimization is very sensitive to local curvature and can easily get stuck on the very jagged grid-based potential surface. The quasi-Newton is slightly more accurate when performing continuum calculations, however.
get_parm pdbfile [PDBPARM_file]where pdbfile is the name of receptor PDB file (with hydrogens added!) that was used as input to chemgrid, and PDBPARM_file is the name of the
PDBPARM file written by chemgrid. PDBPARM_file defaults to
PDBPARM if omitted.
parms directory.
REMARK records for each ligand are provided if so specified at run-time.
| program | mode | parameter file to use |
|---|---|---|
| dockmin_sim | continuum | min.cont.sim |
| dockmin_sim | grid-based | min.grid.sim |
| dockmin_dfp | continuum | min.cont.dfp |
| dockmin_dfp | grid-based | min.grid.dfp |
grid_spacing * 10 ^ (-iexpon).
Smaller iexpon values hence increase the step-size and permit greater
ligand movement - results tend to provide greater accuracy (as compared with
continuum minimization results). Larger iexpon values, however, will
increase speed of minimization by limiting amount of movement. Optimal values
for iexpon seem to be in the range of 5 to 10 when using the trilinear
interpolation energy function. Results may not be exactly consistent
with these trends upon small changes in iexpon, due once again to the
jagged potential surface.