Chemistry Reference
In-Depth Information
screen where every active compound is identified as a hit and every inactive
compound falls below the threshold, the ROC curve approaches the top left corner
(
Se
0) (Fig.
6b
).
Hit list diversity is also an important consideration for the success of a virtual
screen since there is more value in identifying a few unique compounds instead of
many compounds all based on the same chemical scaffold. One way that diversity
can be determined is by comparing the structural similarities of hits from a virtual
screen by using the Tanimoto index [
134
] and then clustering the results. Basically,
a Tanimoto index is calculated based on the fraction of similar chemical sub-
structures present in two structures. Generally, 1,365 chemical substructures
are used to describe a structure. The substructures include individual elements,
two-atom substructures, single rings, condensed rings, aromatic rings, other rings,
chains, branches, and functional groups:
¼
1 and 1
Sp
¼
C
TI
¼
C
;
(8)
A
þ
B
þ
where
A
represents the substructural features present in the first structure,
B
repre-
sents the substructural features present in the second structure, and
C
represents
the substructural features common to both structures. Identical structures have
a
TI
score of 1, where completely dissimilar structures have a
TI
value of 0.
4 Combining Molecular Docking with NMR Ligand
Affinity Screens
The vast majority of initial leads in drug discovery are identified from HTS
[
13
,
135
,
136
]. Pharmaceutical companies have invested heavily in developing
and maintaining large chemical libraries (
1,000,000 compounds), which are
screened using automated, biological assays intended to monitor a specific response
or biological effect [
136
]. Unfortunately, HTS is extremely inefficient due to the
high cost of developing, maintaining, and screening such large libraries of
compounds. Furthermore, the random search for an effective drug in the vastness
of chemical space (~10
60
compounds) [
137
] is almost guaranteed to fail. Thus,
HTS hit rates are typically very low, where
>
0.5% of compounds exhibit any
inhibitor activity in an assay [
138
]. Correspondingly, HTS assays are highly ineffi-
cient since most of the screening effort is spent on the analysis of negative data.
Additionally, HTS assays, by nature, are mechanistic “black boxes,” and a response
does not provide any information on the mechanism of inhibition. This often leads
to numerous false positives from undesirable interactions [
11
,
12
,
139
] that may
lead the drug discovery project astray. Improving the efficiency of drug discovery
requires the implementation of advanced techniques that better guide the selection
of lead candidates without sacrificing speed.
<
