tackle the target flexibility issues in SBDD. Dynamic pharmacophore models recognize compounds, which complementarily bind to the protein considering flexibility of their binding pockets, theoretically reducing the entropic penalties
experienced by the protein due to ligand binding. MD simulation trajectories would
give rise to multiple conformations of a protein active site, describing the targets’
intrinsic flexibilities. Multiple copy minimization is also a regularly used exercise in
computational drug design. The technique first fills the active sites of the receptors
with multiple copies of probe molecules those do not react among themselves.
Then, molecular dynamics, Monte Carlo/steepest descent minimizations are performed to minimize all these probes parallelly to obtain local minima. When the
probes are clustered in the various regions of the active site in different orientations,
the relative preferences of the binding regions can be estimated from the number of
probes or the interaction energies.
Highly ordered and smaller clusters represent highly crucial prerequisite for
favourable interactions, while the haphazardly spread larger ones indicate highly
flexible sites. The MUSIC algorithm [80, 90] implemented with the BOSS programme uses similar strategy. It is capable of performing Monte Carlo simulations
for a wide range of biomolecular systems in solvent clusters and mixtures and
periodic solvent boxes with multiple solutes. It is able to calculate the interaction
energies between solvent–solvent, solvent–solute and solute–solute. Usually, the
probe or solvent are small molecules. For example, hydroxyl groups, aromatic
groups and carbonyl groups are represented by small probes like –CH 3 OH, C 6 H 6
(Benzene) and –CH 3 CO (acetone), respectively. The probe molecules as well as the
side chains of the receptor can be treated as rigid, partially/fully flexible or all-atom.
The wide-ranging OPLS force field used in this programme is proven to be successfully handling the flexibilities of the receptor while generating pharmacophore
models. Applications of the dynamic pharmacophore models will be discussed in
the subsequent sections of the chapter.
6 Pharmacophore Finger Prints
The complex 3D structure of a molecule is reduced to an abstract collection of
features in the pharmacophoric approach. Extending this concept, the structure of a
molecule can be interpreted a as an exclusive data string by extracting all possible
three-/four-point sets of pharmacophoric features. The inter-feature distances are
assigned using distance binning or simply by bonds. These resulting unique strings
describing the frequency of every possible combination at predefined loci of the
string are known as pharmacophore fingerprints. Different types of molecular
similarity analyses among libraries of molecules have been carried out using
pharmacophore fingerprints [91, 92]. Also, the pharmacophoric fingerprint can be
used to detect the common key features/groups contributing to the biological
function of a group of active ligands.
Pharmacophore Modelling and Screening: Concepts, Recent …
39
experienced by the protein due to ligand binding. MD simulation trajectories would
give rise to multiple conformations of a protein active site, describing the targets’
intrinsic flexibilities. Multiple copy minimization is also a regularly used exercise in
computational drug design. The technique first fills the active sites of the receptors
with multiple copies of probe molecules those do not react among themselves.
Then, molecular dynamics, Monte Carlo/steepest descent minimizations are performed to minimize all these probes parallelly to obtain local minima. When the
probes are clustered in the various regions of the active site in different orientations,
the relative preferences of the binding regions can be estimated from the number of
probes or the interaction energies.
Highly ordered and smaller clusters represent highly crucial prerequisite for
favourable interactions, while the haphazardly spread larger ones indicate highly
flexible sites. The MUSIC algorithm [80, 90] implemented with the BOSS programme uses similar strategy. It is capable of performing Monte Carlo simulations
for a wide range of biomolecular systems in solvent clusters and mixtures and
periodic solvent boxes with multiple solutes. It is able to calculate the interaction
energies between solvent–solvent, solvent–solute and solute–solute. Usually, the
probe or solvent are small molecules. For example, hydroxyl groups, aromatic
groups and carbonyl groups are represented by small probes like –CH 3 OH, C 6 H 6
(Benzene) and –CH 3 CO (acetone), respectively. The probe molecules as well as the
side chains of the receptor can be treated as rigid, partially/fully flexible or all-atom.
The wide-ranging OPLS force field used in this programme is proven to be successfully handling the flexibilities of the receptor while generating pharmacophore
models. Applications of the dynamic pharmacophore models will be discussed in
the subsequent sections of the chapter.
6 Pharmacophore Finger Prints
The complex 3D structure of a molecule is reduced to an abstract collection of
features in the pharmacophoric approach. Extending this concept, the structure of a
molecule can be interpreted a as an exclusive data string by extracting all possible
three-/four-point sets of pharmacophoric features. The inter-feature distances are
assigned using distance binning or simply by bonds. These resulting unique strings
describing the frequency of every possible combination at predefined loci of the
string are known as pharmacophore fingerprints. Different types of molecular
similarity analyses among libraries of molecules have been carried out using
pharmacophore fingerprints [91, 92]. Also, the pharmacophoric fingerprint can be
used to detect the common key features/groups contributing to the biological
function of a group of active ligands.
Pharmacophore Modelling and Screening: Concepts, Recent …
39