Chemistry Reference
In-Depth Information
A key question is what is the optimum number of fragments that are required to make
up an effective fragment screening library. Chemical space for fragment molecules is
expected to be considerably smaller than the chemical space of molecules with drug-
like physical properties, so a fragment library should sample fragment chemical space
more efficiently than a drug-like library of the same size can sample drug-like chem-
ical space. It is hard to obtain a good fix on the true size of druggable chemical space
as there is significant variation in the estimates reported in the literature. [ 8 ] Recently,
Reymond and co-workers have built a database of all possible organic molecules with
up to 11 atoms from a restricted set of C, N, O and F. [ 9, 10 ] This contains 26.4 million
compounds (110.9 million stereoisomers) with an average molecular weight (MW) of
153 Da. It is extrapolated that there should be approximately 10 27 different molecules at
25 atoms, which is the size of average drug molecules (MW 340), since the number of
molecules in the Reymond database increases exponentially with the square of the num-
ber of atoms. [ 10 ] It has been further estimated by Guida and co-workers that the universe
of organic compounds containing up to 30 C, N, O and S atoms is in excess of 10 60
different molecules. [ 11 ] Based on analyses such as these, it has been argued that the uni-
verse of potential molecules is very large and it is by no means certain that any screening
library will adequately sample this chemical space. The universe of possible molecules
decreases significantly in size with decreasing molecular weight and so even a relatively
small fragment library will do a better job of sampling fragment chemical space than a
very large screening library samples drug-like chemical space. Indeed, Hann et al . have
argued that lower molecular weight molecules exhibit reduced complexity compared with
the molecules in drug-like collections and have developed a model to rationalise ligand-
receptor interactions in the molecular recognition process. [ 12 ] According to this analysis,
it was shown that the probability of observing a useful interaction falls dramatically with
increasing molecular complexity of the ligand. It can be argued that drug-like collections
such as typical HTS screening decks contain molecules that are too complex to yield use-
ful binding events. Several analyses have shown that as leads are optimised into drugs,
certain properties, such as MW and log P , almost always increase during this process. [ 13, 14 ]
It would therefore be more efficient to start by screening molecules that are more lead-like
rather than a set of molecules that have property profiles more akin to mature, optimised
drug molecules.
Akey consideration in fragment-based drug discovery [ 15 ] is the construction of a fragment
library that is fit for purpose in relation to the types of targets to be screened and the screening
method itself. As with the construction of HTS libraries, [ 16 18 ] there are different approaches
that can be taken and philosophies to be followed. [ 19 ] However, it is usually a matter of
conjecture as to whether one approach for assembling a screening library is superior to
another. [ 20 ] The ideal outcome from an HTS campaign is multiple series of hit compounds
with robust, reproducible activity, target specificity (i.e. not promiscuous inhibitors), novel
and diverse with respect to any known chemical matter for the target of interest, providing
some initial structure-activity relationship (SAR) and suitable for rapid optimisation. The
ideal outcome from fragment screening is broadly similar with the additional requirement
that the active fragments should provide sufficient ligand efficiency and give rise to X-ray
and/or NMR structures in complex with the target protein to merit taking forward into
fragment-to-lead (F2L) optimisation. However, the number of fragment molecules that are
currently available from commercial sources is considerably smaller than the number of
Search WWH ::




Custom Search