Prediction of Protein-Protein Binding Interfaces - Identification of Ligand Binding Site and Protein-Protein Interaction Area - page 105

take advantage of the Rosetta modeling package to predefine docking sites and design
a refinement pipeline which consists of global and local search steps.
Other toolkits available on the Internet but not detailed in this chapter include
Autodock 3 (Cosconati et al. 2010 ), eHiTs (Zsoldos et al. 2006 ) , MOE (Feldman
and Labute 2010 ), FlexX (Kramer et al. 1999 ), ClusPro (Kozakov et al. 2010 ) ,
GRAMM-X (Tovchigrechko and Vakser 2006 ), the PatchDock and SymmDock
(Schneidman-Duhovny et al. 2005 ) programs based on shape complementarity
principles and symmetry restrictions, as well as Hex which bases on spherical
harmonic representations (Macindoe et al. 2010 )

6.2 Programs Description

The models applied are presented in alphabetic order.

6.2.1 Fuzzy Oil Drop Model

The “fuzzy oil drop” model aims to not only recognize protein complexation sites,
but also discover the mechanisms which cause proteins to form complexes (Konieczny
et al. 2006 ). Having explained the premise of the model in the previous chapter we
will now limit ourselves to a brief recapitulation of its key features. At the core of the
model lies the assumption that proteins which undergo folding in an aqueous envi-
ronment tend to internalize hydrophobic residues while exposing hydrophilic resi-
dues on their surfaces. Entropic considerations suggest that a standalone protein
molecule should assume a globular form as a result of interaction with water. Once
folded, the protein possesses a clearly identifiable hydrophobic core (hence the refer-
ence to Kauzmann's “oil drop” concept (Kauzmann 1959 )), which can be modeled
with a 3D Gauss-like hydrophobicity distribution field. Particularly good agreement
with this model can be observed e.g. in fast-folding proteins (Roterman et al. 2011a,
b ), although it should be noted that most proteins exhibit certain deviations from the
“ideal” hydrophobicity distribution. Such deviations are caused by the influence of
external factors on the folding process - this includes ligands and other protein
molecules which form complexes with the protein in question.
The

Δ H profile is a measure of the discrepancy between the expected hydropho-
bicity (given by Gauss' distribution - Ht ) and the actual (observed) hydrophobicity
for the i th aminoacid (or, more specifically, for its effective atom, placed at the geo-
metric center of the amino acid's side chain). Actual hydrophobicity ( Ho ) can be
determined by calculating hydrophobic interactions between the amino acid and all
of its neighbors in a 9 Å radius. According to Levitt (Levitt 1976 ) :

i

Δ= −

HHt Ho

i i