Imperial College London

Henry Rzepa

**[FAIR data]{style="color: #ff0000;"}** is increasingly accepted as a
description of what research data should aspire to;
**[F]{style="color: #ff0000;"}**indable,
**[A]{style="color: #ff0000;"}**ccessible,
**[I]{style="color: #ff0000;"}**nter-operable and
[**R**]{style="color: #ff0000;"}e-usable, with
[C]{style="color: #ff0000;"}ontext added by rich metadata (and also that
it should be[ O]{style="color: #ff0000;"}pen). But there are two sides
to data, one of which is the raw data emerging from say an instrument or
software simulations and the other in which some kind of model is
applied to produce semi- or even fully processed/interpreted data. Here
I illustrate a new example of how both kinds of data can be made to
co-exist.

I will start with a recent
publication\[cite\]10.1002/anie.201407165\[/cite\] with the
title *Crystallographic Snapshot of an Arrested Intermediate in the
Biomimetic Activation of CO~2~. *The nature of this intermediate caught
the eye of another research group, who responded with their own
critique\[cite\]10.1002/anie.201411654\[/cite\]^‡^ along with the
comment "*However, since we have no access to the original
crystallographic data ..." *They might have been referring to the
semi-processed data (containing the so-called *hkl* structure factors)
but they may also have been alluding to the **raw image data** captured
directly from the diffractometer cameras. That traditionally has not
been available via the CSD (Cambridge structural database), but would be
required for a complete re-analysis of the crystal structure. Now the
first example of how both FAIR (processed) data and raw data can
co-exist has appeared.

The latest version of the CSD database shows an entry resulting from the
following publication\[cite\]10.1021/acsomega.7b00482\[/cite\] and the
deposited data has its own DOI there
([10.5517/ccdc.csd.cc1n9ppb](https://doi.org/10.5517/ccdc.csd.cc1n9ppb)).
That entry in turn has a DOI pointer to the **Raw data**
([10.14469/hpc/2300](https://doi.org/10.14469/hpc/2300)) held in a
different location and the pointer is reciprocated (⇌) with the latter
pointing back to the former. Both datasets point to the original
article, thus completing a holy triangle.^†^

![](http://www.ch.ic.ac.uk/rzepa/blog/wp-content/uploads/2017/12/214-1024x995.jpg){.aligncenter
.size-large .wp-image-19184 decoding="async" width="540"
srcset="https://www.ch.ic.ac.uk/rzepa/blog/wp-content/uploads/2017/12/214-1024x995.jpg 1024w, https://www.ch.ic.ac.uk/rzepa/blog/wp-content/uploads/2017/12/214-300x292.jpg 300w, https://www.ch.ic.ac.uk/rzepa/blog/wp-content/uploads/2017/12/214-768x746.jpg 768w, https://www.ch.ic.ac.uk/rzepa/blog/wp-content/uploads/2017/12/214.jpg 1786w"
sizes="(max-width: 1024px) 100vw, 1024px"}

There is more. The Raw
dataset ([10.14469/hpc/2300](https://doi.org/10.14469/hpc/2300))
declares it is a member of a superset, called *Crystal structure data
for Synthesis and Reactions of Benzannulated Spiroaminals;
Tetrahydrospirobiquinolines
(*[10.14469/hpc/2297](https://data.hpc.imperial.ac.uk/resolve/?doi=2297&access=)*) *where
you can find information about six other related structures. That
collection is in turn a member of a superset called *Synthesis and
Reactions of Benzannulated Spiroaminals; Tetrahydrospirobiquinolines
(*[10.14469/hpc/2099](https://doi.org/10.14469/hpc/2099)*) *where DOIs
to other types of data associated with this project can be found, such
as *Computational data*
([10.14469/hpc/2098](https://data.hpc.imperial.ac.uk/resolve/?doi=2098&access=))
and *NMR data*
([10.14469/hpc/2294](https://data.hpc.imperial.ac.uk/resolve/?doi=2294&access=)).
Although a human can with some determination follow these associations
up, down and across, the system is designed to also be followed by
automated algorithms that could traverse this web quickly and
efficiently.

So you can now see that a crystal structure held in the CSD could be the
starting point for a journey of FAIR data discovery, in manner that has
not hitherto been possible. How quickly the CSD will become populated by
links to **Raw** (and other) **data** remains to be seen. I have not yet
discovered any mechanism for specifying a CSD query which stipulates
that Raw data must be available, but no doubt this will come.

To end, back to the *Biomimetic Activation of CO~2~* referred to at the
start. With no access to the original data, recourse was made to
computational modelling.\[cite\]10.1002/anie.201411654\[/cite\] Which
where  I came in, since I wanted access to the original (computational)
data. Sadly it did not appear to be available with the
article,\[cite\]10.1002/anie.201411654\[/cite\] in much the same manner
as the original complaint. Perhaps, when FAIR data becomes fully
accepted as part of how science is done nowadays, such complaints will
become ever rarer!

------------------------------------------------------------------------

^‡^In fact the original authors did
respond\[cite\]10.1002/anie.201504197\[/cite\] with an acknowledgement
that their original conclusions were not correct.

^†^Almost. The article \[cite\]10.1021/acsomega.7b00482\[/cite\] cites
DOI: [10.14469/hpc/2099](https://doi.org/10.14469/hpc/2099) (Ref
28), but it does not cite
DOI: [10.5517/ccdc.csd.cc1n9ppb](https://doi.org/10.5517/ccdc.csd.cc1n9ppb)
because the latter had not been minted yet at the time the final proofs
were corrected, and there is no mechanism to add it at a later stage.

::: {.wp_orcid_field}
[https://orcid.org/0000-0002-8635-8390](https://orcid.org/0000-0002-8635-8390){target="_blank"
rel="author"}
:::


<strong>
 FAIR data
</strong>
is increasingly accepted as a description of what research data should aspire to; **F**indable, **A**ccessible, **I**nter-operable and
<strong>
 R
</strong>
e-usable, with Context added by rich metadata (and also that it should be Open). But there are two sides to data, one of which is the raw data emerging from say an instrument or software simulations and the other in which some kind of model is applied to produce semi-

Messaggi di Rogue Scholar

FAIR data ⇌ Raw data.