ThereĆare several programmes available for detection of clefts, crevices and
binding pockets and to suggest possible active site locations based on the geometry
of the surface [78, 79]. The key residues can be determined by user, deduced from
studying the activity of the protein after mutation of a single residue. If mutation of
a particular residue hampers function of the protein, then that residue may be part of
the active site. Computational analyses such as multiple protein structural alignment
techniques also help in identifying the active site of a protein by comparing it with a
similar protein with known active site.
5.2.2 Complementary Image Construction
The receptor binding pocket is analysed to create an interaction map of features that
the molecule is anticipated to satisfy for a reasonable interaction with the active site.
In other words, a complement of the receptor binding site is created as the basis to
create an input pharmacophore model. In particular, functional features like HB
donors/acceptors and hydrophobic groups are identified in the binding site followed
by rational placement of complementary features within the binding pockets in
chemically acceptable positions [80, 81].
5.2.3 Generation of Queries, Searching and Hit Analysis
Once the active site is defined and chemically characterized, there is no straightforward single step to derive pharmacophore models from the binding site
map. Since the receptor binding site has a potential to bind a variety of molecules in
a variety of binding conformations, the interaction map often gives rise to huge
number of features. To address this problem, adjacent features of the same type are
clustered and the feature that lies nearest to the geometric centre of the cluster is
retained as the cluster representative and all the other features are discarded.
Sometimes, the number of the features is still very high even after the clustering,
and all of them cannot be used as a single model because models possessing all
such features would not be able to obtain any hits from the database. So, possible
combinations of limited numbers of features are derived from the interaction map
and multiple pharmacophore modes are composed. And then, these models are used
by programmes like catalyst [43, 82] implemented in Accelrys Discovery Studio to
search the compound database and test the validity of the models (also termed as
pharmacophore ‘queries’ in catalyst) to screen or reject highly active compounds. It
is always necessary to examine these models for how they interact with the binding
site residues and how far the models extend within the binding pocket and if they
fill specificity pockets and make the strongest interactions. Queries describing only
the features present in an inhibitor might end up giving many false positive hits. At
times, they screen compounds that are able to map to all the query features but also
contain a bulky substituent causing steric hinderance and averting the compound
36
C. Choudhury and G. Narahari Sastry
binding pockets and to suggest possible active site locations based on the geometry
of the surface [78, 79]. The key residues can be determined by user, deduced from
studying the activity of the protein after mutation of a single residue. If mutation of
a particular residue hampers function of the protein, then that residue may be part of
the active site. Computational analyses such as multiple protein structural alignment
techniques also help in identifying the active site of a protein by comparing it with a
similar protein with known active site.
5.2.2 Complementary Image Construction
The receptor binding pocket is analysed to create an interaction map of features that
the molecule is anticipated to satisfy for a reasonable interaction with the active site.
In other words, a complement of the receptor binding site is created as the basis to
create an input pharmacophore model. In particular, functional features like HB
donors/acceptors and hydrophobic groups are identified in the binding site followed
by rational placement of complementary features within the binding pockets in
chemically acceptable positions [80, 81].
5.2.3 Generation of Queries, Searching and Hit Analysis
Once the active site is defined and chemically characterized, there is no straightforward single step to derive pharmacophore models from the binding site
map. Since the receptor binding site has a potential to bind a variety of molecules in
a variety of binding conformations, the interaction map often gives rise to huge
number of features. To address this problem, adjacent features of the same type are
clustered and the feature that lies nearest to the geometric centre of the cluster is
retained as the cluster representative and all the other features are discarded.
Sometimes, the number of the features is still very high even after the clustering,
and all of them cannot be used as a single model because models possessing all
such features would not be able to obtain any hits from the database. So, possible
combinations of limited numbers of features are derived from the interaction map
and multiple pharmacophore modes are composed. And then, these models are used
by programmes like catalyst [43, 82] implemented in Accelrys Discovery Studio to
search the compound database and test the validity of the models (also termed as
pharmacophore ‘queries’ in catalyst) to screen or reject highly active compounds. It
is always necessary to examine these models for how they interact with the binding
site residues and how far the models extend within the binding pocket and if they
fill specificity pockets and make the strongest interactions. Queries describing only
the features present in an inhibitor might end up giving many false positive hits. At
times, they screen compounds that are able to map to all the query features but also
contain a bulky substituent causing steric hinderance and averting the compound
36
C. Choudhury and G. Narahari Sastry