validation tools and services were first introduced around two decades ago, there was
a tendency to use them as a binary gauge of quality, i.e. if a structure fails to meet all
test criteria, then it is simply not acceptable. While this attitude can still be encountered, on the whole it has become standard practice to use this service as a tool to
assist review and complement a reviewer’s own judgement – that is, to use it in
combination with other assessment approaches. These combined approaches to
assess ‘validity’ of a crystal structure have become increasingly important as the
fields of chemistry and structural chemistry have evolved over the last few years.
Crystallographers are posed with increasingly difficult problems, where often it is
not possible to recrystallise a product but crystallographic characterisation is crucial.
We are beginning to move to a culture, supported by the right processes and tools,
where the notion that ‘data needs to be fit for the purpose’ is acceptable. In some
areas of chemistry, any amount of structural information, e.g. connectivity or
conformation, can provide huge insights and be immensely valuable, whereas in
certain studies differences of a thousandth of an angstrom for a particular interaction
can be very important. Clearly one needs to be able to make a judgement as to
whether the evidence/data that supports a claim is of an appropriate level of quality
to do so. We now have in place most of the standards, tools and processes to be able
to do this. So long as the appropriate approaches are taken, it should now be perfectly
acceptable to publish a structure that traditionally would have had an unacceptably
high R-factor, so far as the structure is basically ‘correct’ and the author is not
attempting to make grossly overstated claims in their analysis.
As a result, publishers are now only one aspect of the research lifecycle where
quality is assessed. Database providers now perform more quality assessment before
accepting deposits and as part of the process of validating structures for entry into
their collections, but also end users now have the tools to deduce these quality levels
themselves. These factors mean that now it should be possible to rapidly make
crystallographic data available and if the right processes are followed and correct
tools applied, then the end user can readily make use of this data with the confidence
that it is being used appropriately. This is particularly important as we see the rise of
computational/theoretical (a very significant proportion of the users of crystal
structure data are performing in-silico calculations on them) and cheminformatics
approaches in research – especially areas that are purely consumers of crystal
structures.
However, we do need to be careful! If one considers that the crystal structure
results we are talking about are in fact just models and based on how raw data has
been interpreted, then there may well be cases where simply assessing and validating
a regular result, i.e. a CIF, is not enough.
The small molecule crystallography community, through the IUCr CommDat, is
now assessing in great detail the extent to which the raw data underpinning the CIF
result needs also to be made available [64]. There are strong arguments and trends
that this will happen as a routine matter of course in the macromolecular crystallography community; however, there are valid reasons why this might not be the right
approach for small molecule crystallographers. Firstly, the sheer volume of results in
small molecule crystallography is much greater, and so ‘publishing’ associated raw
Leading Edge Chemical Crystallography Service Provision and Its Impact on. . .
95
a tendency to use them as a binary gauge of quality, i.e. if a structure fails to meet all
test criteria, then it is simply not acceptable. While this attitude can still be encountered, on the whole it has become standard practice to use this service as a tool to
assist review and complement a reviewer’s own judgement – that is, to use it in
combination with other assessment approaches. These combined approaches to
assess ‘validity’ of a crystal structure have become increasingly important as the
fields of chemistry and structural chemistry have evolved over the last few years.
Crystallographers are posed with increasingly difficult problems, where often it is
not possible to recrystallise a product but crystallographic characterisation is crucial.
We are beginning to move to a culture, supported by the right processes and tools,
where the notion that ‘data needs to be fit for the purpose’ is acceptable. In some
areas of chemistry, any amount of structural information, e.g. connectivity or
conformation, can provide huge insights and be immensely valuable, whereas in
certain studies differences of a thousandth of an angstrom for a particular interaction
can be very important. Clearly one needs to be able to make a judgement as to
whether the evidence/data that supports a claim is of an appropriate level of quality
to do so. We now have in place most of the standards, tools and processes to be able
to do this. So long as the appropriate approaches are taken, it should now be perfectly
acceptable to publish a structure that traditionally would have had an unacceptably
high R-factor, so far as the structure is basically ‘correct’ and the author is not
attempting to make grossly overstated claims in their analysis.
As a result, publishers are now only one aspect of the research lifecycle where
quality is assessed. Database providers now perform more quality assessment before
accepting deposits and as part of the process of validating structures for entry into
their collections, but also end users now have the tools to deduce these quality levels
themselves. These factors mean that now it should be possible to rapidly make
crystallographic data available and if the right processes are followed and correct
tools applied, then the end user can readily make use of this data with the confidence
that it is being used appropriately. This is particularly important as we see the rise of
computational/theoretical (a very significant proportion of the users of crystal
structure data are performing in-silico calculations on them) and cheminformatics
approaches in research – especially areas that are purely consumers of crystal
structures.
However, we do need to be careful! If one considers that the crystal structure
results we are talking about are in fact just models and based on how raw data has
been interpreted, then there may well be cases where simply assessing and validating
a regular result, i.e. a CIF, is not enough.
The small molecule crystallography community, through the IUCr CommDat, is
now assessing in great detail the extent to which the raw data underpinning the CIF
result needs also to be made available [64]. There are strong arguments and trends
that this will happen as a routine matter of course in the macromolecular crystallography community; however, there are valid reasons why this might not be the right
approach for small molecule crystallographers. Firstly, the sheer volume of results in
small molecule crystallography is much greater, and so ‘publishing’ associated raw
Leading Edge Chemical Crystallography Service Provision and Its Impact on. . .
95