23 / 25
Crystallography 2018
Structural Chemistry & Crystallography Communication
ISSN: 2470-9905
Page 48
June 04-05, 2018
London, UK
3
rd
Edition of International Conference on
Advanced Spectroscopy,
Crystallography and Applications
in Modern Chemistry
T
hecrystalstructuresimilarityisnotanestablishedandgenerally
defined property. There are various definitions of crystal
structure similarity defined for different purposes, each having
different advantages and disadvantages in different situations.
There are several methods, that define the similarity of crystal
structures as a similarity of a representative functions called
fingerprints [1]–[4]. These methods compare crystal structures
indirectly, by comparison of their fingerprints. Other methods, are
trying to compare atomic coordinates [5]–[8] or even positions
of basic moieties in the crystal structures [9], [10]. In all cases,
when differences in positions of atoms or moieties are used for
calculation of the crystal structure similarity, the transformation
between crystal structures has to be determined. The difficulty
of this procedure is nicely described in [8]. CrystalCMP [11] is a
software for comparison of molecular packing that was recently
published. The suggested method is based on the second
mentioned approach - comparison of molecular positions. It is
immediately clear, that the comparison method is designed for
all non-polymeric crystal structures, where some stand-alone
moieties (molecular fragments) can be found. It is perfectly
valid for all molecular crystals and some of the metal-organic
complexes. Most of the inorganic structures and MOF with
polymeric structures cannot be compared by this method. The
comparisonmethod is divided in several steps: (i) Definition of the
central molecule (the largest molecule in the unit cell by default),
(ii) creating of the molecular cluster (10 surrounding molecules
by default), which is representing the whole crystal structure,
(iii) definition of the fragment for overlaying (either by SMILES
notation or by HASH strings as originally published in 2016)
and (iv) overlapping molecular clusters according to the defined
fragment and (v) calculating differences in molecular positions
and its relative rotations, see definition of the Psab formula.
where Dc is the average distance (in Å) between the molecular
centers of related molecular pairs and Ad is the average angle
(in degrees) between them. The X value is set by the user to
weight the influence of the Ad parameter (the default value is X
= 100), see Fig 1. As a result of comparison is a similarity matrix
with calculated dendrogram and the transformation matrix
between both compared molecular clusters. This enables
overlaying the compared structures and see differences
visually in human-readable form. The advantage of this
method is its low sensitivity to the relatively large expansion
of the molecular structure caused e.g. by the temperature or
even by the presence of different solvent molecules in the
crystal structure. For that reason this method is applicable
for comparison of solvatomorphic series of identical or even
just similar compounds. Several tests on different compounds
had been performed. The algorithm compares two molecular
packing in less than one second on a common office PC
(approx. 100 ms for small molecule of benzamide and approx.
200 ms for middle-size molecule of trospium [11]). This allows
making comparison of large number of compounds. In addition,
automation of the method allows, for example, comparison of
all crystal structures in the whole CSD database
Figure 1 A graphical interpretation of the method used in CrystalCMP. (Top left
and top right) In both cases the central molecules are surrounded by eight mole-
cules. (Bottom) Both upper clusters are overlapped and the numbers near each
surrounded molecule represent the differences in the molecular centers of re-
lated pairs and the angle differences between them. In general, these numbers
are real. (Left) Two crystal structures with almost identical packing, Psab = 0.5
+ 100 × (2.5/180) = 1.9 (for X = 100). (Right) Two crystal structures with almost
identical positions of surrounding molecules, but with different packing of sur-
rounding molecules, Psab = 0.5 + 100 × (90/180) = 50.5 (for X = 100).
Recent Publications
1. M. Valle and A. R. Oganov, “Crystal fingerprint space – a
novel paradigm for studying crystal-structure sets,” Acta
Crystallogr. A, vol. 66, no. 5, pp. 507–517, Sep. 2010.
2. E. L. Willighagen, R. Wehrens, P. Verwer, R. de Gelder,
and L. M. C. Buydens, “Method for the computational
CRYSTALCMP – FAST PACKING COMPARISON OF MOLECULAR CRYSTALS
Jan Rohlicek
Institute of Physics, ASCR, Czech Republic
Jan Rohlicek, Struct Chem Crystallogr Commun 2018, Volume 4
DOI: 10.21767/2470-9905-C1-005