Chemistry Reference
In-Depth Information
the targets against which it will be tested will not have been determined. Indeed, in our
experience, screening collections are challenged by a diverse range of targets during their
useful lifetime and on this basis it can be beneficial to have a set of chemically diverse
molecules at the heart of the library that can be augmented with focused or hypothesis-
driven sets of molecules on a target-by-target basis. We advocate that a screening collection
should be regarded as a dynamic rather than a static entity.
3.2.1
Sources of Molecules
Before selecting molecules for a fragment library, a pool of candidate molecules will be
required for which there are several possible sources. As outlined in the introduction,
numerical estimates for the possible number of molecules within the drug-like portion
of chemical space are vast.
[
8 11
]
To help with a further classification of chemical space,
Hann and Oprea introduced a concept of the virtual, the tangible and the real worlds of
molecules.
[
27
]
The virtual world consists of molecules that are possible in theory, the tangible
world includes molecules that are chemically feasible (and the number of molecules within
its realm will increase as new chemistries become available) and, finally, the real world
contains molecules that are physically available. In principle, the selection of fragment
molecules can be made from the tangible and real worlds of molecules. Once the molecular
structures are available for tangible compounds, these can be processed in much the same
way as real world molecules, subject to the caveat that any molecules that are selected can
in fact be synthesised. In practical terms, the tangible and real worlds of molecules will
vary to some extent between different organisations, with some having access to novel
chemistry, synthetic routes and building blocks, etc., in addition to each having access to
their own internal compound collections.
Commercially availablemolecules occupy both the tangible and real worlds ofmolecules.
Many organisations maintain copies of these vendor catalogues in internal databases which
can be searched and filtered as required. Evotec maintains a database of commercial screen-
ing compounds which is compiled from the currently available collections of over 35
different compound vendors and is updated on a regular basis. After removal of duplic-
ate structures, the database contains over 3.8 million chemical structures and is therefore
representative of the real world of commercially available screening compounds. As such,
this database provides a useful resource for the selection of fragment molecules in library
enhancement initiatives and the generation of focused sets, and also for the selection of
molecules in fragment hit follow-up activities.
3.2.2 Physical Property and Substructure Filters
Physical property and substructure filters are straightforward to apply to large databases and
are often used as one of the initial steps in a fragment selection process. Fragment molecules
are sometimes defined in terms of their physical properties, which in turn can be used to help
shortlist molecules in fragment selection processes. In a similar manner to Lipinski's rule
of five for the identification of molecules with poor absorption or permeability,
[
6
]
Congreve
et al
. proposed a rule of three for fragment-based lead discovery.
[
7
]
The criteria for the
rule of three are shown in Table 3.1 and were deduced from the analysis of a diverse set of
fragment hits that were obtained against a range of targets. Hann and Oprea have designed a
