Biomedical Engineering Reference
In-Depth Information
1.6 MOLECULAR DIVERSITY AND CHEMICAL SPACE
Chemical space, or more properly, multidimensional descriptor space , encompasses
all theoretically possible compounds and is therefore essentially infinite, limited
only by the imagination of chemists and current synthetic methodologies [37,38].
Molecules occupy discrete points within chemical space with “similar” molecules
grouped together and “dissimilar” molecules farther apart. Molecules' positions
in chemical space are determined by their comparable physical properties, such as
molecular weight, log P, and polarizability as well as their topological features [37,39].
An algorithm based on a large number of these descriptors can be used to create a
representation of chemical space based on the descriptors used and the limits placed
on them. A molecule's position within this particular multidimensional descriptor
space can then be calculated. To give a visually accessible representation of multi-
dimensional descriptor space, it is necessary to use principal components analysis
(PCA) [40] to condense the information into two- or three-dimensional scatter plots.
These plots then provide a means to easily compare the relative coverage of mul-
tidimensional descriptor space achieved by compound collections. It is, however,
worth mentioning that these plots are not absolute assessments of diversity or chem-
ical space coverage, as there is the potential for a molecule's relative and absolute
position to move depending on the molecular descriptors chosen and any weighting
scheme applied to the analysis [41]. Because of this potential, scatter plots are often
produced with two or more compound collections superimposed on each other so that
their relative diversity can be compared. Figure 1.3 shows an example of chemical
space analysis produced using chemical descriptors and PCA.
1.7 SYNTHETIC STRATEGIES FOR CREATING
MOLECULAR DIVERSITY
As noted earlier, the challenge of creating molecular diversity efficiently is a consid-
erable one, requiring strategies that differ from the majority of traditional chemical
syntheses. Since the beginnings of DOS in the early 2000s, two distinct strategies
for the generation of molecular diversity (in particular, skeletal diversity) have been
identified in the literature [5]. They are: (1) a reagent-based approach , where sub-
jecting a given molecule to a range of reaction conditions allows the synthesis of a
number of distinct compounds; and (2) a substrate-based approach , where a num-
ber of starting materials containing preencoded skeletal information are transformed
under the same conditions into a range of molecular structures (Figure 1.4). These
The words similar and dissimilar are used with caution, as these terms require a point of reference
against which to compare. As such, the same set of molecules could be considered similar or dissimilar,
depending on how they are compared (the descriptors used). However, within a given analysis more
“similar” molecules should group closer together than those with traits that are more different.
Search WWH ::




Custom Search