Biomedical Engineering Reference
In-Depth Information
specific goals of the study and the resources available. Molecular properties, struc-
ture fingerprints, and molecular scaffolds are major types of structure representation
that are used for quantitative study of the structure diversity and chemical space of
compound databases such as LOLs and other collections obtained from DOS, natural
products, compounds annotated with biological activity, marketed drugs, commer-
cial vendor libraries, and virtual collections. These comparisons have provided basic
information used to guide the selection, design, and synthesis of compound libraries
for improved drug discovery. For example, frequent scaffolds identified in bioac-
tive compounds and scaffolds present in natural products represent starting points to
inspire the diversity-oriented synthesis of biologically relevant and natural product-
like libraries that contain compounds with increased probability of having biological
activity. Chemoinformatic analyses of various types of compound databases have
shown that the medicinally relevant chemical space can be biased toward the currently
known chemical organic universe (i.e., the organic compounds synthesized thus far).
This observation is supported by several studies of scaffold content analyses of cur-
rently available compound collections, which indicate that just small portions of syn-
thetically available scaffolds have actually been made. As such, the pharmaceutically
relevant space may be greater than it is known currently, due to the still limited knowl-
edge of pharmaceutically relevant targets and the complex relationships revealed by
polypharmacology studies. To address this problem, increasing the molecular com-
plexity of screening collections while balancing the physicochemical properties has
been proposed as a feasible strategy for expanding the medicinally relevant chemical
space. In this regard, DOS approaches such as LOLs represent promising strategies
for exploring systematically neglected chemical and scaffold spaces.
Despite the fact that computational analysis of the chemical space and structure
diversity of compound collections is conducted by industrial, academic, nonprofit, and
other research groups, it is not a simple task. Characterizing chemical spaces poses
challenges that the research community is pursuing actively. Innovative computational
approaches are being developed in response to novel findings and emerging needs of
the drug discovery community. For example, the large impact of ADME attributes in
the clinical success of bioactive compounds has increased the awareness to employ
ADME-related properties to visualize and mine ADME spaces. Although it has been
conceptually attractive to dissect the biologically relevant chemical space in subspaces
containing islands of compounds directed to specific therapeutic indications, the con-
cept of polypharmacology challenges this idea. The constantly increasing chemoge-
nomics data have stimulated the development of computational approaches to mine
the multidimensional biological space of compound data sets and has led to novel
concepts such as chemotography. The well-known dependence of chemical space
with structure representation has boosted the development of approaches to make a
consistent comparison of the chemical space across research groups. In this chapter
we have illustrated recent efforts toward a consistent framework to represent the
chemical space as demonstrated by the development of delimited reference chemi-
cal spaces, molecular quantum numbers, ChemGPS, and ligand-receptor interaction
fingerprints.
