Biomedical Engineering Reference
In-Depth Information
description of the chemical compounds using a set of characteristics such as molec-
ular mass, lipophilicity, and topological features to name a few, Dobson conceptual-
ized the chemical space as “the total descriptor space that encompasses all the small
carbon-based molecules that could in principle be created” [30]. The chemical space
is vast, as shown by the calculations of the total number of possible compounds that
could be made (e.g., 10
60
) [31]. As discussed in subsequent sections, one of the major
applications of chemical space is the comparison of compound data sets to design
novel collections and/or select existing compound libraries for experimental or com-
putational screening. A second major application of the notion of chemical space
is the classification of bioactive compounds according to their biological properties.
The latter application is based on the hypothesis that similar molecules have similar
bioactivity [32,33]. Therefore, it is expected that, at least in principle, the “biologi-
cally active space” [34] would be formed by separated clusters of compounds, each
one associated with a different receptor [29]. The latter concept has led to the field
of biology-oriented synthesis (BIOS), which aims to target “bioactivity islands” with
compound data sets containing core structures of compound classes that are relevant
to nature [12,35].
10.3 GENERAL ASPECTS OF CHEMOINFORMATIC METHODS TO
ANALYZE THE CHEMICAL SPACE
Chemoinformatics
, also referred in the literature as
cheminformatics
or
chemical
information science
has been defined as “the application of informatic methods to
solve chemical problems” [36] or as “a scientific field based on the representation of
molecules as objects (graphs or vectors) in a chemical space” [37]. Additional defi-
nitions are reviewed by Varnek and Baskin [37] and Willett [38]. Chemoinformatics
plays a key role in the diversity analysis of compound collections and the mining
of chemical space. Major aspects of chemoinformatics include the representation of
chemical compounds, storing and mining information in databases, and generating
and analyzing data [36].
Based on the definition of chemical space by Dobson as discussed above, Le
Guilloux et al. pointed out the need to define chemical spaces for chemoinformatic
applications using a restricted number of compounds and number of descriptors [39].
To this end, a large number of molecular representations, such as physicochemi-
cal properties, fingerprint-based representations, and molecular scaffolds, have been
used to describe chemical spaces and to compare compound databases. The use
of a comprehensive description of the molecules often leads to multidimensional
spaces that require dimension-reduction techniques to visualize such multidimen-
sional arrays and extract useful information. In fact, molecular representation and
data-reduction techniques are two major factors that have a strong influence on the
analysis and visual representation of chemical spaces [40,41]. A main aspect of
dimension-reduction techniques is to maintain a large percentage of the information
from higher-dimensional space [34,42]. Different sets of descriptors and parameters
used to define the space where the molecules will be located [39] may change the
