Chemistry Reference
In-Depth Information
9
Virtual Fragment Scanning:
Current Trends, Applications and
Web-based Tools
Bradley Feuston, M. Katharine Holloway, Georgia McGaughey
and J. Christopher Culberson
9.1
Introduction
Significant resources are currently devoted to screening fragment databases with the goal of
impacting lead finding and optimization in drug discovery. To facilitate this approach, it is
clear that molecular modeling plays an important role in interpreting results and providing
insights to help guide subsequent efforts.
While there are numerous references delineating the successes of fragment-based screen-
ing through technologies such as NMR, X-ray and surface plasmon resonance (SPR), which
are discussed elsewhere in this topic, the present chapter focuses on various aspects of virtual
screening of fragment libraries. Crystal structures of proteins with bound small fragments
have in general identified a number of conformations and binding sites for each ligand.
[
1
]
Such a lack of specificity in binding modes for fragments presents a difficult challenge for
virtual screening. From a modeling perspective, small fragments are much weaker bind-
ers than typical drug-like molecules and are more challenging to score.
[
2
]
To normalize
the binding energy of ligands by size, the concepts of binding efficiency index and ligand
efficiency have evolved to choose preferentially weak binders with low molecular weight
or low non-hydrogen atom counts, respectively.
[
3, 4
]
The same problem exists with virtual
screening and scoring functions. A universal scoring function for bound ligands remains a
