by other approximations. These advantages should not be under-
estimated. Not only are they important from an epistemological
perspective (they provide results and explanations), they also pro-
vide a visual support to their predictions and visual support is
extremely important to communicate results to nonspecialized
audiences (e.g., scientific collaborators from other fields, investors).
Having said so, one should have in mind that the efficacy of a given
technique is highly dependent on the chosen molecular target.
Regarding VS approaches, a gold standard has not been found
yet, a fact that explains the need of rigorous in silico validation
before moving to VS and subsequent wet experiments. Some vali-
dation approaches are briefly discussed (seeNote 4).
Frequently, different techniques are complementary in nature
[39] and the simplest methods have surprisingly good outcomes in
some cases. This allows the definition of hybrid protocols combin-
ing simple and complex approximations either serially or in parallel
[40] (Fig.3); serial combined approaches tend to provide robust
solutions.
A final and important step to prune the hits emerging from
systematic screening involves filtering out promiscuous com-
pounds, unspecific inhibitors and reactive compounds, such as
PAINS and REOS filters [41, 42].
Fig. 3While in parallel VS hybrid methods result in combination of complemen-
tary sets of hits (thus retrieving more chemical diversity), serial hybrid methods
tend to produce more robust, consensus hit sets
Computer-Aided Drug Design 9