Pharmacophore and FTrees
Abhik Seal
IUPAC Definition: “An ensemble of steric and electronic features that is necessary to ensure
the optimal supramolecular interactions with a specific biological target and to trigger (or
block) its biological response“
In drug design, the term 'pharmacophore‘ refers to a set of features that is common to a series
of active molecules
Hydrogen-bond donors and acceptors, positively and negatively charged groups, and
hydrophobic regions are typical features
We will refer to such features as 'pharmacophoric groups'
• Bioisosteres, which are atoms, functional groups or molecules with similar
physical and chemical properties such that they produce generally similar
biological properties .
• A chemical group can be mimicked by a similar group with
similar biological activity –another example of similarity
for example in
a. Size
b. Shape (bond angles, hybridization)
c. Electronic distribution (Polarizability,
inductive effects, charge, dipoles)
e. Lipid solubility
f. pKa
g. Chemical reactivity (including
likelihood of metabolism)
h. Hydrogen bonding capacity
3D Pharmacophores
A three-dimensional pharmacophore specifies the spatial relation-ships between the groups
Expressed as distance ranges,angles and planes
A commonly used 3D pharmacophore for antihistamines contains two aromatic rings and a
tertiary nitrogen
Tak Taken from Laak etal.
J.Med Chem1995,38(17)
Example of ACE inhibitors..
Angiotension-converting enzyme (ACE), which is involved in regulating blood
pressure .
Interacts with an Arg
residue of enzyme
H bonds to a hydrogen-bond donor in enzyme
a zinc-binding group
4 points 5 distances
Detection of Pharmacophores:
A pharmacophore software detects the elements which is responsible for pharmacophoric
For Ligand based pharmacophore pharmacophoric points.
a) Aromacity detection or ring detection
b) HBD point is normaly bsaed on topological information .Every atom is checked for the
following conditions:
Only nitrogen or oxygen atoms;
• Formal charge is not negative;
• At least 1 attached hydrogen atom.
c) The generation of hydrogen bond acceptor points needs to fulfill four conditions:
• Only nitrogen or oxygen atoms;
• Formal charge not positive;
• At least one available lone pair;
• Atom is ‘accessible’.
d) For the generation of charge center pharmacophore points, the formal charges on the atoms
of the molecule are used. Atoms with a positive formal charge will correspond with a positive
charge center pharmacophore point
Structure based pharmacophore
Distance Constraints represent the relation between two points, one located on the ligand side, one
on the macromolecular side.The following table shows LigandScout's default distance constraint settings:
Aromatic interaction with
positive ionizable
3.5 - 5.5 Å
Aromatic interaction with ring
2.8 - 4.5 Å
Aromatic interaction with ring
2.8 - 4.5 Å
H-Bond interaction
2.2 - 3.8 Å
Hydrophobic interaction
1.0 - 5.9 Å
Iron binding location
1.3 - 3.5 Å
Magnesium binding location
1.5 - 3.8 Å
Negative ionizable interaction
1.5 - 5.5 Å
Positive ionizable interaction
with negative ionizable
1.5 - 5.5 Å
Positive ionizable interaction
with aromatic ring
1.0 - 10.0 Å
Zinc binding location
1.0 - 4.0 Å
Merging and aligning Pharmacophores
The quantification of the similarity between two pharmacophores can be computed from the
overlap volume of the Gaussian volumes of the respective pharmacophores.
The procedure to compute the volume overlap between two pharmacophores is
implemented in two steps.
a) a list of all feasible combinations of overlapping pharmacophore points is generated.
b) then corresponding features are aligned with each other using an optimization algorithm.
The combination of features that gives the maximal volume overlap is retained to give the
matching score
Each pharmacophore point is modeled as a 3-dimensional spherical Gaussian volume
represented by its center (coordinate) and spread (a). The definition of the Gaussian
volume is given as follows:
Vp 
 pexp m  r
Vp being the Gaussian volume, p being normalization constant to scale the total
volume to a level that is in relation to atomic volumes, m being the center of the
Gaussian, and r being
 the distance variable that is integrated.  that defines the
volume of the point in space.  is chosen inverse proportional to the square root
of the radius.
3D database searching
4 - 7.2 Å
The first stage employs some form of rapid screen to eliminate molecules
that could not match the query.
For eg: One way to develop is to encode information of the distances in the form
of Bit strings. Where each bit position would correspond to a range of distance between
specific pair of atoms. For initialization at first the bit string is set to 0 at all bit positions and
then for each molecular conformation the bit string positions are set to 1. Then the final
Encoded bit string is searched against a database to look for similar molecules.
The second stage uses a graph-matching procedure to identify those structures that do truly
match the query.
Eg : Clique detection.
Clique Detection methods
When many pharmacophoric groups are present in the molecule it may be very
difficult to identify all possible combinations of the functional groups
Clique is defined as a 'maximal completely connected subgraph'
Clique detection algorithms can be applied to a set of pre-calculated conformations
of the molecules
Cliques are based upon the graph-theoretical approach to molecular structure .
similar pattern
• Molecular descriptors are used for retrieval of compounds
and also for clustering and and property prediction.
• Most descriptors use today are in linear format such as the
properties are stored in the form of a vector.
• The alignment free approach of comparison is extremely
fast but it has disadvantages i.e the relative arragnment of
funtional groups on the molecular surface cannot be
determined and its weakly described in linear descriptors.
• On the other hand 3D model can be itself considered as
descriptor and they are aligned in 3D space , but its is
difficult due to conformational flexibility and it might miss
the right alignment.
Feature Trees
• Mixed 2D and 3D ligand-based approach
• Alignment based but conformation independent descriptor.
• A feature tree represents a molecule by a tree such that the
tree should capture the major Building blocks of the molecule in addition to
the overall alignment.
• In this way lead hopping between Chemical classes with compounds
Sharing the same wanted biological activity is supported.
The nodes of the Ftrees represents the fragments
of the molecule.
Each atom of the molecule is associated with with at least one node.
• Two nodes which have atoms in common or which contain atoms connected in
the molecular graph are connected.
• The feature tree nodes are marked with labels
describing the shape and chemical properrties of the
bulding block.
Taken from Rarey etal JCADD 1998
Descriptors in Ftrees
• The shape descriptors has two components i.e the number of
atoms and the approximated vander wall’s volume.
• Chemical features is used to describe the interaction pattern
the Building Ftree can form The FlexX interaction pattern is
used which is represented .
• All the features taken are additive
Taken from Rarey etal JCADD 1998
Comparison algorithm
• The comparison algorithm of two feature trees is based on matching the
trees i.e a subtree of one feature to that of another.
• a,b is the number of atoms of the compared fragments.i th entry describes
the ability of a Fragment to form interaction of type i.
• On comparing the full Ftree we just add
all the features and apply the eq. such a
Comparison is level -o
For comparing feature Trees split search and match
Search algorithms are used .These algorithms
match based on the topology I,e it maintains the
Scaffold hopping example
• The Ftrees can identify actives from decoys Of H4R receptor proteins.
Score 0.835
Thank you.

similar documents