drug repositioning [103–105]. Firstly, pharmacophore-based fingerprints can be
employed to search for similar molecules, whose mechanisms of action are already
understood. In the other way around, pharmacophore models can be generated from
the active sites of a group of probable proteins involved in the particular disease
pathway and then the active molecules can be mapped to them to find out the best
fit. The structures of these groups of proteins may be obtained from PDB or models
generated using various techniques. The active site pharmacophore mapped with
high scores can be proposed as potential targets for the compounds. A study on a
group of plant metabolites and pharmacophore models of their possible targets was
carried out by Rollinger et al. The best mapping targets were later proven to be
accurate by experimental testing, thus validating the usefulness of the pharmacophore mapping approach [106].
7.4 De Novo Ligand Design with Pharmacophores
Apart from acting as a query to screen molecules with features at desired spatial
locations and thus possibly prompting a desired biological response, pharmacophore models can also be employed for de novo design, of compounds, satisfying
a specific physicochemical constrains. For example, the NEWLEAD method is able
to create novel molecules from distinct disconnected fragments (mostly derived
from known active ligands) that are consistent with the features of a pharmacophore
model by using linkers. The linkers are small connecting fragment may be few
atoms, chains or sometimes ring moieties [107]. Software packages like LUDI
[108] or BUILDER [109] can grow such novel molecules when the receptor
structures are also known. Many other packages also perform such de novo ligand
design from the receptor-based pharmacophore features [110, 111]. Thus, pharmacophore models have versatile ways of application for lead generation. De novo
design is meant to create entirely novel compounds, while pharmacophore
searching screens the available chemical space. However, pharmacophore searching is faster and easier.
8 Limitations of Pharmacophore-Based Approaches
Though the literature is flooded a plenty of successful and reliable applications of
pharmacophore-based approaches in rational drug design, its limitations should be
cautiously considered as with any method [33, 112]. A systematic or straightforward way of constructing pharmacophore models is not available. This is the case
especially with the receptor-based pharmacophore models where many different
combinations of features are possible and each model may screen completely different set of molecules [113]. Lack of accuracy in pharmacophore scoring/fitness
functions is one of the limitations of pharmacophore searching. So, quality of
Pharmacophore Modelling and Screening: Concepts, Recent …
47
employed to search for similar molecules, whose mechanisms of action are already
understood. In the other way around, pharmacophore models can be generated from
the active sites of a group of probable proteins involved in the particular disease
pathway and then the active molecules can be mapped to them to find out the best
fit. The structures of these groups of proteins may be obtained from PDB or models
generated using various techniques. The active site pharmacophore mapped with
high scores can be proposed as potential targets for the compounds. A study on a
group of plant metabolites and pharmacophore models of their possible targets was
carried out by Rollinger et al. The best mapping targets were later proven to be
accurate by experimental testing, thus validating the usefulness of the pharmacophore mapping approach [106].
7.4 De Novo Ligand Design with Pharmacophores
Apart from acting as a query to screen molecules with features at desired spatial
locations and thus possibly prompting a desired biological response, pharmacophore models can also be employed for de novo design, of compounds, satisfying
a specific physicochemical constrains. For example, the NEWLEAD method is able
to create novel molecules from distinct disconnected fragments (mostly derived
from known active ligands) that are consistent with the features of a pharmacophore
model by using linkers. The linkers are small connecting fragment may be few
atoms, chains or sometimes ring moieties [107]. Software packages like LUDI
[108] or BUILDER [109] can grow such novel molecules when the receptor
structures are also known. Many other packages also perform such de novo ligand
design from the receptor-based pharmacophore features [110, 111]. Thus, pharmacophore models have versatile ways of application for lead generation. De novo
design is meant to create entirely novel compounds, while pharmacophore
searching screens the available chemical space. However, pharmacophore searching is faster and easier.
8 Limitations of Pharmacophore-Based Approaches
Though the literature is flooded a plenty of successful and reliable applications of
pharmacophore-based approaches in rational drug design, its limitations should be
cautiously considered as with any method [33, 112]. A systematic or straightforward way of constructing pharmacophore models is not available. This is the case
especially with the receptor-based pharmacophore models where many different
combinations of features are possible and each model may screen completely different set of molecules [113]. Lack of accuracy in pharmacophore scoring/fitness
functions is one of the limitations of pharmacophore searching. So, quality of
Pharmacophore Modelling and Screening: Concepts, Recent …
47