Biomedical Engineering Reference
In-Depth Information
[137]. These descriptors include atom, bond, polarity, and topology counts. TheMQN
space of PubChem, the entire chemical universe database GDB-13 (with 977 million
molecules up to 13 atoms of C, N, O, S and Cl), and DrugBank have been developed
and are freely available [138,139].
In an independent effort to represent the biologically relevant chemical space
as a general reference, Rabal and Oyarzabal recently developed a novel descrip-
tor called a ligand-receptor interaction fingerprint (LiRIf). This descriptor converts
structural information into a one-dimensional string accounting for the plausible
ligand-receptor interactions as well as for topological information, and it is useful
to cluster, profile, and compare chemical libraries [140]. This work represents a sig-
nificant advance toward the development of generally applicable representations of
chemical space for which no reference space is required [140].
As discussed above, there is an increasing amount of biological information of
compound data sets that have been screened across multiple biological endpoints. To
analyze the SAR of complex chemogenomics data, the concept of chemotography
(chemotype chromatography), has been proposed as an approach that combines visu-
alization of chemical space with analysis of rich biological data [141]. The approach
implemented by Lounkine et al. allows an easy comparison of different representa-
tions of chemical space [141].
Recent advances in scaffold analysis include the development of tree maps, which
provide a novel way to visualize the distribution of molecules over scaffolds, and
the molecular similarity of the scaffolds within a compound data set [107]. In that
work, Langdon et al. used tree maps to visualize the scaffold space of drugs, in-
house screening collections, focused libraries, compounds from ChEMBL, and other
collections. Tree maps provide an easy way to display highly populated scaffolds and
cluster similar scaffolds [107].
An activity landscape has been conceptualized as a chemical space with the
addition of biological activity as another dimension [142]. This concept, which is
at the interface between chemical space and biological space, has emerged as an
attractive approach to analyze the SAR of compound collections systematically and
detect
activity cliffs
(chemical compounds with highly similar structures but signif-
icantly different biological activities) [40] and
scaffold hops
(chemical compounds
with highly dissimilar structures but significantly similar biological activities) [92].
Although a comprehensive review of activity landscapes is beyond the scope of this
chapter, the interested reader is referred to Chapter 16 and to extensive reviews of
progress in this field [126,142-144].
10.6 CONCLUDING REMARKS
Chemoinformatic characterization of the chemical space and molecular diversity of
compound libraries is a central topic in drug discovery and other research areas.
Unlike the cosmic universe, chemical space is not unique and it is highly influ-
enced by the molecular representation. The criterion to describe the chemical space
is a critical factor in deriving useful conclusions, and it is frequently based on the
