|
Optimal Performances
UVSS is a highly scalable software that increases the speed of your
docking engines : when controled by UVSS, your docking tools can benefit
from fast local I/Os, and the distribution of results is parallelized
by UVSS across multiple NFS servers: no I/O bottleneck means
less processors and less docking software licences costs for the same
throughput. UVSS increases the ROI.
The high throughput virtual screening process
has been modelized by Genomining to get optimal values for the parallelization parameters.
For instance, the optimal size of each group of ligands that are
docked simultaneously or in a single local node has been estimated, expressed in terms
of several parameters (see figure).
Two types of parameters will affect the execution time of a pipeline :
-
configuration parameters that affect the settings of the docking engines (ex.: choice
of the scoring function, number
of iterations of the search algorithm, number of conformations per ligand...)
-
distribution policy: size of each docking group.
The correctness of the results will be affected by the first type of
parameters but it will be independant of the distribution policy.
Hence, the user can select an optimal value for the group size
parameter before submitting a pipe. This optimal value, Sopt, is the
group size associated with the global minimum value of the following curve~:
|
