Biomedical Engineering Reference
In-Depth Information
that have similar activity. Hence, in LBVS, known active compounds are used as refer-
ence molecules. In principle, such molecules are not required for SBVS, although the
knowledge of active compounds also provides additional information for structure-
based approaches (e.g., known actives can be modeled into binding sites to guide
compound selection). It is important to note that SBVS and LBVS approaches are
not mutually exclusive. In many cases, especially for popular pharmaceutical tar-
gets, both target structure and ligand information is available and SBVS and LBVS
can be applied in parallel or in a coordinated manner. Furthermore, SBVS efforts
greatly benefit from the availability of x-ray structures of ligand-target complexes,
which makes it possible to study compound binding modes directly and use this
information to guide docking studies. This knowledge usually greatly improves the
likelihood of identifying novel active compounds. However, SBVS is in principle a
de novo approach to hit identification, whereas LBVS is always comparative in nature:
that is, it attempts to relate chemical and structural features of known active com-
pounds and candidate molecules to each other as a basis for predictions. Accordingly,
in LBVS, the question of ligand novelty plays a central role.
15.2.2 Scaffold Analysis
One is always interested in identifying hits that depart structurally from already
known actives: those that contain previously unobserved core structures (also called
scaffolds
) [18]. Therefore, the primary LBVS goal is generally referred to as
scaffold
hopping
[19,20]. By contrast, one is usually not interested in identifying molecules
that are closely related to or are structural analogs of known active compounds,
because such compounds can be selected “by eye” or by substructure searching,
without the need to apply VS methods. However, the assessment of scaffold hopping
potential in LBVS is generally complicated and often biased by several factors. For
example, chemically distinct scaffolds in compounds having similar activity might
span a wide structural range from very similar ones (which might be distinguished
only by individual heteroatom positions and/or bond orders) to essentially unrelated
ones. Thus, scaffold hops often represent very different degrees of chemical distance.
Identifying active compounds with similar scaffolds is not very challenging; however,
finding virtually unrelated scaffolds that represent compounds with similar activity
is rather challenging. It has generally been difficult to compare the scaffold hopping
potential of different LBVS methods (i.e., to assess the chemical distances that dis-
tinguish newly identified hits from reference molecules). Recently, however, a first
distance function has been introduced to quantify the chemical distance between scaf-
folds [21]. By applying this function, it is possible to evaluate and compare numer-
ically the degree of difficulty involved in scaffold hopping exercises. Figure 15.1
shows exemplary scaffolds hops and reports the interscaffold distances.
15.2.3 Methodological Complexity
It is often assumed that methodological complexity in LBVS scales with search per-
formance and scaffold hopping potential. However, this is an unsubstantiated and
