Biomedical Engineering Reference
In-Depth Information
de-prioritize them for further analysis). Clearly, the identification of key compounds
in large data sets (including HTS data) that reveal SAR determinants is a major
attraction of activity landscape exploration. Although activity landscapes are first
and foremost descriptive in nature and do not represent predictive models, compound
design suggestions can be deduced from their analysis, which meets the primary
interest of practicing medicinal chemists.
Different landscape designs convey SAR information in different ways, but activ-
ity cliffs are generally regarded as their most interesting features. However, this
assumption also requires consideration. The study of activity cliffs reveals substi-
tution sites that are critical for compound potency but does not necessarily suggest
which compounds to make next. To understand how structural modifications alter
compound potency in a defined and predictable manner, it is also required to explore
SAR continuity in the vicinity of cliffs. Thus, although high SAR information con-
tent is associated with small structural changes of active compounds leading to large
biological effects, a thorough rationalization of SARs requires consideration of this
informationwithin its chemical context. This is often (but not always) straightforward.
For example, inspecting the environment of prominent activity cliffs might reveal the
formation of coordinated cliffs involving multiple compounds, which provides more
information about SAR determinants than do series of isolated cliffs. Furthermore,
this might reveal patterns of SAR continuity and discontinuity within the same or
similar series of compounds from which SAR determinants might also be deduced.
Activity cliffs are often discussed in a rather loose manner, which makes it difficult
to judge a given analysis or compare results from different studies. Currently, there
are no generally accepted criteria for the description and rationalization of activity
cliffs. However, regardless of whether one considers activity cliffs as a continuum
or as discrete states, it is essential to define them clearly for any computational or
chemical investigation.
On a final note, it is well appreciated that lead optimization requires the con-
sideration of multiple properties beyond compound potency. Hence, if the study of
activity landscapes and cliffs should indeed become an integral component of chemi-
cal optimization efforts, it would ultimately be required to further extend the activity
landscape concept and include multiple chemical and/or biological properties. First
attempts in this direction are being made, and there is considerable potential for future
developments.
Acknowledgment
The author would like to thank Dagmar Stumpfe for reviewing the manuscript and
helping with illustrations.
REFERENCES
1. A. M. Wassermann, M. Wawer, J. Bajorath,
J. Med. Chem.
2010
,
53
, 8209-8223.
2. G. M. Maggiora,
J. Chem. Inf. Model.
2006
,
46
, 1535-1535.
