Current State-of-the-art for Virtual Screening and Docking Methods - New Developments in Medicinal Chemistry - page 91

We subsequently review 50 docking/virtual screening protocols.
GLIDE originally developed by Friesner et al. is now available in the Schrodinger
software suite (Schrodinger LLC) as a module. With geometries and properties of
the binding site receptor, the program generates grids before starting the docking job.
An exhaustive sampling in the ligand molecule torsional space is done to generate
ligand-binding poses during docking which can be divided into four steps. The
program employs, during steps one and two, a series of hierarchical filters to search
for possible locations of the ligand on the previously prepared grids. This is done in
order to generate plausible ligand binding poses via a coarse screening. The steric
complementarities of the ligand to the defined binding sites are examined by the
initial filters evaluating ligand receptor interactions with GlideScore. The ligand
binding poses selected by the initial screening are, in the third stage, minimized with
force fields in situ . A composite score is used at the last stage to rank the resulting
ligand binding poses and select the ones to report by GlideScore. Internal steric
energies as well as nonbonded interaction energies of the generated ligand binding
poses are taken into consideration [492].
GOLD is another option for molecular docking and virtual screening. The method
originally developed by Jones et al. is now released by Cambridge
Crystallographic Data Center. Gold uses a genetic algorithm in order to explore
the conformational space of the ligand taking into account the conformational
flexibility of several selected amino acids residues of the protein. An initial
population of ligand binding poses are randomly generated from the given three-
dimensional structures of protein and ligand. Each individual of the population is
assigned a fitness score based on its predicted binding affinity using implemented
scoring functions (GoldScore, ASP, ChemScore). Ranking of all individuals is
done according to their fitness scores. The entire population is optimized
iteratively via migration operations crossover and mutations. Parameters control
the genetic algorithm sampling procedure (Selection Pressure, Population Size,
Number of Islands, Number of Operations, Niche Size, Migrate, Operator
Weights, Crossover, Mutate) [493].
LigandFit, originally developed by Venkatschalam et al . is now implemented in
the Discovery Studio software (Accelrys Software) as a module. The procedure