Biomedical Engineering Reference
In-Depth Information
MACCS keys
TGD
PCA 2
PCA 2
PCA 1
PCA 1
PCA 3
PCA 3
FIGURE 10.9
Dependence of chemical space with structure representation. Visual repre-
sentation of the chemical space of a total of 2250 compounds selected from nine compound
collections (250 compounds per data set). The data sets are natural products (green) and a
general screening collection (magenta) from two commercial providers, five in-house com-
binatorial libraries (red, yellow, gray, black, and cyan), approved drugs (blue), and a set of
compounds targeted to adenosine receptors (orange). The depiction of the chemical space was
obtained by PCA of the similarity matrix computed by Tanimoto similarity and two different
structure representations. The first three PCs account for 85.9% (MACCS) and 90.3% (TGD)
of the variance, respectively. (
See insert for color representation of the figure
.)
To perform comprehensive comparisons of compound data sets, the use not only
of different fingerprint representations but also of different criteria, such as physic-
ochemical properties, structure fingerprints, and molecular scaffolds, has been pro-
posed [105]. The basis of this approach is that each method has advantages and
disadvantages. For example, the use of molecule properties has the advantage of
being intuitive and straightforward to interpret. As discussed above, the drug-like
[58] and lead-like criteria [59], and Congreve's rule of 3 [60] have been formu-
lated using physicochemical properties. However, physicochemical properties do not
provide information regarding the structural patterns, and molecules with different
chemical structures can have the same or similar physicochemical properties. Also,
chemotypes or scaffolds are straightforward to interpret by computational and medic-
inal chemists as well as biologists. For example, scaffold analysis has led to concepts
such as scaffold hopping [130] and privileged structures [131,132]. However, one of
the disadvantages of the scaffold analysis is the lack of information regarding struc-
tural similarity due to the side chains and the inherent similarity or dissimilarity of the
scaffolds themselves. An obvious solution is the analysis not only of the molecular
frameworks but also of the side chains, functional groups, and other substructural
analysis strategies [133]. To consider the entire molecular structure in the compar-
isons, molecular fingerprints are widely used and have been applied to a number
of chemoinformatic and computer-aided drug design applications [32,124,134]. Fol-
lowing the three criteria of physicochemical properties, structure fingerprints, and
molecular scaffolds, a collection of natural products implemented in the ZINC
