Chemistry Reference
In-Depth Information
P1
P2
P3
P4
P5
Structural
superimposition
Docking
Ligand
Optimization
Min(E(x,y,z,
θ
,
φ
,
ψ
,m))
Ligand
P3
Fig. 5 A cartoon illustration of ensemble docking, where five individual protein structures are
superimposed to create a single scoring parameter for the docked ligand. Ensemble docking
minimizes the computational effort since a single docking occurs to select the best conformer
instead of five separate molecular docking simulations. (Reprinted with permission from [
118
],
copyright 2007 by John Wiley and Sons)
A virtual screen never results in all the truly active compounds being top ranked.
Instead, most virtual screening protocols set a binding score or ranking threshold to
identify the predicted active compounds or “hits.” In general, top ranked compounds
are expected to be enriched with active compounds compared to a random selection
(Fig.
6a
). A high enrichment factor (
EF
>
10) is considered the benchmark of success
for a virtual screening [
132
]. Enrichment is dependent on sensitivity (
Se
)andspecific-
ity (
Sp
). Sensitivity represents the true positive rate, which is the ratio of true positives
(
TP
) found by the virtual screening vs the total number of actives (
A
) in the library. The
number of actives corresponds to both true positive (
TP
) and false negative (
FN
):
TP
Se
¼
FN
:
(5)
TP
þ