3.4 Areas Where Data-Driven Methods Are Making
an Impact
The CSD can not only be used to search for structures and evaluate new structures
but can also be used as a diverse experimental test set for computational algorithms,
which seek to learn trends from the data and therefore predict molecules along with
their useful properties. Machine learning approaches have received growing interest
in recent years. Applications of the increasing volume of data in the CSD have been
extensively covered in the recent review article by Taylor and Wood [15]. The single
biggest research area where applications of the CSD are used is the pharmaceutical
industry; however, there are many other areas where the database has been used to
further research. The following section highlights a few examples of areas where
crystallographic data from the CSD has been used to aid research.
3.4.1 Pharmaceuticals
The drug discovery journey begins with the search to find a molecule that has the
correct shape and intermolecular interactions to bind in the target active site of a
protein. The CSD can be used not only to search for suitable molecules but can also
be exploited as a database of experimental conformations. These geometric preferences can be utilised when generating potential molecules in determining the most
likely conformations [145, 169]. Intermolecular interaction data from the CSD is
used in evaluating the interactions between the ligand and the binding site, especially
to identify where changes to the molecule can be made to optimise the binding or
physical properties of the molecule. Two approaches to drug candidate optimisation
are bioisosteres, chemical substituents or groups with similar physical or chemical
properties which produce broadly similar biological properties, and scaffold hopping, the search for compounds containing different core structures which retain
similar activity due to the 3D location of binding groups [171–173].
In order to be able to develop an identified active molecule into a marketed drug,
it is important to understand the solid form that will be given to the patient. This
requires investigating a range of solid form characteristics, such as if the molecule is
polymorphic or could have solvate or hydrate forms. While many techniques are
utilised, the crystal structure plays an invaluable role in assessing a compound. The
CSD has had a key role in underpinning collaborations with industry in the development of tools and methods to address these questions, firstly with the Pfizer
Institute for Pharmaceutical Materials Sciences and then from 2008 the Crystal
Form Consortium. A paper by Galek et al., celebrating the half millionth structure
added to the CSD, is an example of the type of informatics that could be employed
[174]. An in-depth analysis of the intermolecular interactions using Full Interaction
Maps and Hydrogen Bond Propensity enabled an assessment of whether there are
any unusual hydrogen bonds or unsatisfied hydrogen bond donor and acceptors in
the crystal form. The presence of these may indicate the possibility of a more stable
Leading Edge Chemical Crystallography Service Provision and Its Impact on. . .
119
an Impact
The CSD can not only be used to search for structures and evaluate new structures
but can also be used as a diverse experimental test set for computational algorithms,
which seek to learn trends from the data and therefore predict molecules along with
their useful properties. Machine learning approaches have received growing interest
in recent years. Applications of the increasing volume of data in the CSD have been
extensively covered in the recent review article by Taylor and Wood [15]. The single
biggest research area where applications of the CSD are used is the pharmaceutical
industry; however, there are many other areas where the database has been used to
further research. The following section highlights a few examples of areas where
crystallographic data from the CSD has been used to aid research.
3.4.1 Pharmaceuticals
The drug discovery journey begins with the search to find a molecule that has the
correct shape and intermolecular interactions to bind in the target active site of a
protein. The CSD can be used not only to search for suitable molecules but can also
be exploited as a database of experimental conformations. These geometric preferences can be utilised when generating potential molecules in determining the most
likely conformations [145, 169]. Intermolecular interaction data from the CSD is
used in evaluating the interactions between the ligand and the binding site, especially
to identify where changes to the molecule can be made to optimise the binding or
physical properties of the molecule. Two approaches to drug candidate optimisation
are bioisosteres, chemical substituents or groups with similar physical or chemical
properties which produce broadly similar biological properties, and scaffold hopping, the search for compounds containing different core structures which retain
similar activity due to the 3D location of binding groups [171–173].
In order to be able to develop an identified active molecule into a marketed drug,
it is important to understand the solid form that will be given to the patient. This
requires investigating a range of solid form characteristics, such as if the molecule is
polymorphic or could have solvate or hydrate forms. While many techniques are
utilised, the crystal structure plays an invaluable role in assessing a compound. The
CSD has had a key role in underpinning collaborations with industry in the development of tools and methods to address these questions, firstly with the Pfizer
Institute for Pharmaceutical Materials Sciences and then from 2008 the Crystal
Form Consortium. A paper by Galek et al., celebrating the half millionth structure
added to the CSD, is an example of the type of informatics that could be employed
[174]. An in-depth analysis of the intermolecular interactions using Full Interaction
Maps and Hydrogen Bond Propensity enabled an assessment of whether there are
any unusual hydrogen bonds or unsatisfied hydrogen bond donor and acceptors in
the crystal form. The presence of these may indicate the possibility of a more stable
Leading Edge Chemical Crystallography Service Provision and Its Impact on. . .
119