Biomedical Engineering Reference
In-Depth Information
been anticipated that the computational and experimental characterization of DOS,
other combinatorial libraries, natural products, and other screening collections will
be conducted to explore their potential to expand the medicinally relevant chemical
space [55].
10.4.3 Scaffold Analysis
The scaffold or molecular framework is used to describe the core structure of a
molecule, and it is one of the concepts most widely used in medicinal chemistry
and drug discovery [92]. Applications of the scaffold content in small data sets or
large compound collections [4] include studying the SAR/SPR of a set of molecules
with measured biological activity or other property; analyzing the structural diver-
sity of compound collections, and evaluating the performance of virtual screening
methodologies to retrieve novel scaffolds [93].
Several scaffold analyses of compound databases have been performed for dif-
ferent purposes [94-105]. For example, Bemis and Murcko analyzed a collection of
drugs to identify common structural features and found that a rather small number of
scaffolds accounts for a large fraction of molecules in the database [94]. This obser-
vation is in line with a more recent work of Langdon et al. that analyzed the scaffold
diversity of approved drugs, compounds from commercial vendors, focused libraries
targeted to kinases, compounds from the ChEMBL database [106], and molecules
from in-house collections. It was found that most of the compounds in the databases
analyzed are contained in a small number of well-represented scaffolds and that there
are a large number of scaffolds with only one compound [107].
Scaffold content analysis of more than 24 million organic compounds in the CAS
Registry showed that a small percentage of frameworks occur in a large percentage
of compounds (e.g., the most common 5.0% of the hetero frameworks were found
in 75.5% of the compounds) [108]. This observation supports the hypothesis that the
currently known medicinally relevant chemical space is biased to the rather small
fraction of compounds that have been made (see above) [35].
With the goal of identifying properties that are associated with biological activity,
the properties and composition of heteroaromatic ring systems that are present in
bioactive molecules has been compared to the properties of a large collection of
virtual ring systems [109]. In that study, Ertl et al. concluded that the number of
unique scaffolds is very small (ca. 0.5% in the databases analyzed) and that the
bioactive aromatics scaffolds form bioactivity islands [109]. Authors suggested that
the most important properties responsible for the separation of active and inactive
areas of chemical space seem to be size of the scaffolds, their heteroatom composition,
and their stability [109]. In an independent study, Pitt et al. showed that there is a
large portion of synthetically accessible aromatic rings that can potentially be made
that has not yet been explored [110].
Scaffold content analysis of compound libraries annotated with biological activity
has given rise to the identification scaffolds frequently found in bioactive compounds.
For example, the authors reported a chemotype-based hierarchical classification of
the AIDS database at the National Cancer Institute in order to identify systematically
