CAMBRIDGE STRUCTURAL DATABASE SYSTEM 2010 RELEASE

WORKED EXAMPLES

Copyright 2009 The Cambridge Crystallographic Data Centre Registered Charity No 800579

Table of Contents

1 2

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Mogul Tutorials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 2.1 Tutorial 1: Validating Molecular Dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 2.2 Tutorial 2: Analysing the ring conformation of a protein bound ligand . . . . . . . . . . . . . . . . 8

Materials Mercury Tutorials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 3.1 Tutorial 1: Searching on Hydrogen-Bond Motifs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 3.2 Tutorial 2: Studying Crystal Packing Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 3.3 Tutorial 3: Searching on Packing Similarity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

IsoStar Tutorials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 4.1 Tutorial 1: Using IsoStar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

1. Introduction This booklet contains a series of worked examples that illustrate a number of application uses of certain software components within the CSD System. The examples are suitable for use as teaching, workshop or training material for those who may be unfamiliar with the Mogul, Materials Mercury, or IsoStar software modules. Mogul Tutorials (see page 2) Materials Mercury Tutorials (see page 14) IsoStar Tutorials (see page 43) Note: User Guides and additional Tutorials for ConQuest, Mercury, Mogul, and IsoStar are available via the Help option in each of the individual interfaces. Alternatively, User Guides and Tutorials for all CSD System software components can be accessed via the CCDC website at the following address: http://www.ccdc.cam.ac.uk/support/documentation

CSDS Worked Examples

2. Mogul Tutorials Mogul is a library of intramolecular geometry and provides instant click of a button access to geometrical distributions of bond lengths, valence angles, torsion angles, and ring conformations. Uses include validation of newly determined crystal structures, identification of unusual geometric features, and checking of conformations generated by computational procedures. Tutorial 1: Validating Molecular Dimensions (see page 3) Tutorial 2: Analysing the ring conformation of a protein bound ligand (see page 8)

CSDS Worked Examples

2.1 Tutorial 1: Validating Molecular Dimensions Introduction Comparing the dimensions of a newly determined small-molecule crystal structure with the bond lengths and angles of similar structures in the CSD is extremely useful both as a check against refinement errors and to highlight unusual geometric features. This tutorial demonstrates how to search on all bond and angle fragments in a query molecule and shows how unusual or even suspect geometric features can be readily identified. Menu Commands Required 1. Import the query structure. Click on the Load button in the Build query pane. In the resulting Load molecule dialogue box, select cyclopropyl.mol2 from <SOFTWARE_INSTALLDIR>\examples\tutorials\ and hit Open. Search on all bond lengths and angles. An All fragments search will allow you to search for all valid bond lengths, valence angles and/ or torsion angles within your query molecule. Click on the All fragments... button on the left of the Build query screen. In the resulting Search for all fragments pop-up, disable the All torsion fragments check-box and ensure that both the All bond fragments and All angle fragments check-boxes are selected. Hit Search to run the search. Viewing the search results. The results from an All-fragments search are displayed (in spreadsheet format) in a separate Allfragments search window.

2.

3.

CSDS Worked Examples

 Results for either bond length or valence angle fragments can be viewed by clicking on the appropriate tab at the top of the All-fragments search window. Click on the Bond tab. Each bond fragment in the query structure is listed in the spreadsheet together with the summary statistics for the corresponding Mogul distribution. 4. Identifying unusual geometric features. For each bond fragment in the query structure statistics are given in the All-fragments search window, these include: number of observations, minimum, maximum, mean, median, standard deviation, value in query and z-score. z-score is the absolute difference between observed and mean values of a geometric parameter divided by the standard deviation of the Mogul distribution. Therefore, a high z-score (e.g. >2.0) may indicate an unusual or even suspect geometry within your query. The rows of a spreadsheet can be sorted according to the values in any of the columns. Click on the z-score column header button to sort the rows by z-score. Notice that the C1-C2 bond length has a high z-score value (2.282). To investigate this potentially suspect bond length further, display the search results for the C1-C2 bond fragment by clicking on the corresponding row in the spreadsheet. In the main Mogul window click on the Build query tab. The C1-C2 bond fragment is directly attached to a cyclopropyl ring and is highlighted in the query structure.

CSDS Worked Examples

5.

Analysing the results Click on the Results and analysis tab in the main Mogul window. The value of the C1-C2 bond length in the query is superimposed in red on the histogram. This allows for easy comparison with the geometric results obtained from Mogul.

CSDS Worked Examples

 When compared to similar structures in the CSD the C1-C2 bond in the query structure appears to be unusually short (1.481 ) and falls outside the main Mogul distribution. However, there are a small number of observations in the histogram with a similar value to that of the query structure. In order to inspect just these structures deselect all hits in the histogram by clicking on the Deselect button, then highlight the three histogram bins located around 1.48 using the horizontal bar located directly under the histogram:

CSDS Worked Examples

 Click on the View structures tab and inspect the CSD entries that contribute to the selected bins. Notice that for each hit structure the search fragment is also attached directly to a cyclopropyl ring. Therefore, it might be reasonable to assume that the shortening of the C1-C2 bond is a consequence of the cyclopropyl group and representative of this type of structural motif (i.e. the C1-C2 bond length in the query structure is in fact correct). In order to confirm this check some of the hit structures in the more populated region of the distribution and satisfy yourself that these do not contain a cyclopropyl group.

CSDS Worked Examples

2.2 Tutorial 2: Analysing the ring conformation of a protein bound ligand The Example The refinement of protein/ligand structures from X-ray diffraction patterns is clearly a harder problem than refinement from small molecule derived diffraction patterns. The resolution of these patterns is lower as a rule and consequently it is very rarely possible to get electron density resolved to the atomic level. It is therefore necessary to assume standard bond lengths and bond angles in a protein refinement and use these to help the refinement process generate reasonable models. Quite frequently however ligand models are generated which are far from being low energy structures. Occasionally this might be because the protein is straining the ligand. More frequently it is because alternative ligand models that might have fitted the electron density equally well, but with lower strain energy, have not been investigated Mogul can be used to examine structures of ligands bound to proteins against similar chemistry within the CSD. An assessment can be made as to whether a ligand structure of unusual geometry is correct and strained by the protein, or alternatively, is unlikely to be correct. A thorough assessment should normally look not only at the ligand structure but also the resolution of the complex, the fit of the ligand structure to the electron density of the complex by visual inspection, and possibly an analysis of the quality and nature of the interactions made by the ligand to the protein. This tutorial illustrates how Mogul can be used to assess two models of a ligand structure from the protein data bank entry 1hak. Menu Commands Required 1. Import the query structure. Click on the Load button in the Build query pane. In the resulting Load molecule dialogue box, select 1hak_ligand_A.mol2 from <SOFTWARE_INSTALLDIR>\examples\tutorials\ and hit Open. Carrying out a Ring search Select all six atoms that make up the piperidine ring (N27 C30 C31 C32 C33 C34) and click
Search.

2.

3.

Analysing the results The histogram showing the distribution of geometries for related rings should appear. The Xaxis gives an average measure of Root Mean Square Difference (RMSD) between each ring torsion in the query and the corresponding ring torsion in each hit.

CSDS Worked Examples

 The search retrieves nine hits with relevance 1.0. All are at the far end of the X-axis at RMSD of 70 degrees indicating that the piperidine geometry in the model is very different from that represented in the CSD. Click on More hits ..., to bring up the Find more hits [ring] window, change the threshold in the Aim for at least box to 45, and click on OK. Does the distribution of ring geometries significantly change? 4. Finding more information: Analysing the torsions. We will now examine whether other geometrical features look odd. Go back to the Build query window and select All fragments. Toggle on All torsion fragments and ensure all the other options re toggled off. Click on Search. The All fragments: Results window that comes up has a column labelled d(min). This measures in degrees the distance of the query torsion to the nearest occupied bin on the histogram. Click on the top of the d(min) column to sort it so that the highest d(min) value is at the top.

CSDS Worked Examples

 Examine on the query structure, the torsions represented by the second to fourth rows (Why should we ignore the first row?). Two torsions are represented, both close to the piperidine fragment. Look also at the histograms for these torsions. Do you think that the query values are reasonable? 5. Looking at a second model structure in 1hak. The structure 1hak has two molecules of protein in the unit cell. Consequently two model structures exist for the ligand bound in the active site. We would normally expect these to exhibit very similar binding mode and geometry although it cannot be ruled out that differences in geometry may really occur, due to real differences in the protein conformation brought about by crystallographic packing and environment. However in many cases a difference in geometry between two available ligand models is due to incorrect ligand model choice, in one or both cases Click on the Load button in the Build query pane. In the resulting Load molecule dialogue box, select 1hak_ligand_B.mol2 from <SOFTWARE_INSTALLDIR>\examples\tutorials\ and hit Open Carrying out a Ring search Select all six atoms that make up the piperidine ring (N27 C30 C31 C32 C33 C34) and click
Search.

6.

7.

Analysing the results The histogram for the ring search indicates that the ring geometry is not particularly unusual. It is in fact close to a chair conformation, which is very common in six membered rings. Perhaps this indicates that this ligand model is a better one?

10

CSDS Worked Examples

 Try to find more hits. Very few hits of relevance 1.0 are found, which might be a significant observation. 8. Finding more information: Analysing the torsions. We will again examine whether other geometrical features look odd. Go back to the Build query window and select All fragments. Toggle on All torsion fragments and ensure all the other options are toggled off. Click on Search. Again sort the All fragments: Results table by d(min). Examine on the query structure the torsions represented by the top three rows. Two torsions are represented, both close to the piperidine fragment. Look also at the histograms for these torsions. The query values are well away from the CSD distribution. Does this then mean that the model is bad after all?

9.

A hypothesis The 1hak_ligand_A.mol2 structure is very likely to be a bad model. The ring conformation is clearly highly unusual. What about the model 1hak_ligand_B.mol2 however? There is a clue to the answer in that very few examples of relevance 1.0 were found for model B. Relevance in rings depends on three factors, number of substituents on each ring atom, size (small or large) of each ring atom substituent; and relative stereochemistry. The piperidine ring
11

CSDS Worked Examples

is substituted only twice, in the 1 and 4 positions and the subtituents themselves are both considered small (because the atom of each substituent adjacent to the ring only has at most one additional heavy atom attached). It is unlikely that other similar examples cannot be found in the CSD, and in fact we know such examples exist because 9 examples with relevance 1.0 were found for the search on 1hak_ligand_A.mol2 . Therefore it must be the relative stereochemistry of the two ring substituents that is the unusual factor, and which forces the ring search to generalise and get other hits. Examination of the piperidine ring in 1hak_ligand_B.mol2 shows both substituents to be UP (or DOWN). Why is this so rare in a 1,4 substituted piperidine? The answer lies in the stereochemical preferences of substituted saturated six membered rings. So called Axial substitutions where the substituent points vertically up perpendicular to a plane through the ring (chair conformation assumed) are less stable (because they make more close contacts to ring Hs) than Equatorial substituents which, as their name suggests, come off the sides of the ring (see diagram). The benzyl group in Model B coming off the a piperidine ring carbon, is in an axial position. The substituent off the nitrogen atom is in the equatorial position. This conformation for the piperidine ring is unusual because the nitrogen in similar compounds is usually able to invert itself (umbrella inversion) so that its substituent is axial. The six membered ring can then also invert by a process called intra-chair conversion and in this process both axial ring substituents can then take up the much more stable equatorial position. Consequently we see that the piperidine ring in Model B still contains significant strain, despite having a common chair conformation, and that this strain is at least in principle resolvable by adopting an alternative conformer.

We see now that both model A and model B are highly strained structures. A full analysis would require us to look at the electron densities in both structures. However, even without that, a working hypothesis might be that the crystallographic refinement of both ligand models is not as
12 CSDS Worked Examples

good as it could be and that, if the crystallographer could have used alternative starting models for each ligand structure, it might have been possible to generate models with good fit to the electron density, which contained low strain piperidine conformations, each a chair form with two equatorial substituents.

CSDS Worked Examples

13

3. Materials Mercury Tutorials Utilising the unique information about crystal packing in the CSD, the Materials module of Mercury allows you to perform novel searches and analyses of the CSD, applicable to problems such as cocrystal design, counter-ion selection, polymorph prediction, and crystal engineering studies. The functionality in the Materials module of Mercury CSD facilitates in-depth analysis and comparison of crystal forms and the study of packing patterns in the CSD. You can: Compare packing patterns of structures and quantify similarity between polymorphs, hydrates and solvates Identify regions of structural similarity Search sets of structures for specific interaction motifs or general packing features and compare geometries Tutorial 1: Searching on Hydrogen-Bond Motifs (see page 15) Tutorial 2: Studying Crystal Packing Features (see page 25) Tutorial 3: Searching on Packing Similarity. (see page 36)

14

CSDS Worked Examples

3.1 Tutorial 1: Searching on Hydrogen-Bond Motifs Introduction Gaining control of how molecules pack in the solid state is a very desirable goal, for many reasons. Hydrogen bond motifs are commonly used as a way of guiding the interactions between molecules. It is important to understand what a likely outcome is for functional group interactions between molecules. We are going to look at the interactions involving the carbamoyl moiety to learn how these groups like to fit together. Menu Commands Required 1. Begin the Motif Search Click on Search in the top-level menu and select Motifs from the resulting menu. In this example, we are going to generate our own motifs, rather than using the pre-defined motifs available in the expandable lists. Select the Creat new motifs radio button and click on Next > to start the Motif Search wizard. 2. Select the functional groups for the motif There are three basic steps in the process of motif generation: Defining functional groups, Selecting contacts and Generating motif search queries. We are going to perform all three. Here you can define functional groups that will be included in the motif search queries. You can either: Add pre-defined groups from a set list using the Add button. Sketch your own groups with a sketcher using the Sketch button. Load groups from saved .con (connser) files using the Load button. The carbamoyl group is a pre-defined functional group so click on Add to choose the group. Select the carbamoyl group from the drop-down list in the Select substructure dialogue

CSDS Worked Examples

15

 click on OK and then click on Next > to proceed to the contacts page. 3. Selecting the contacts Next we have to define the contacts between the functional groups that will be used to generate the motifs. Click on the Add button. In the resulting dialogue, make a contact between the carbamoyl N and the carbamoyl O by selecting the carbamoyl groups from the from and to dropdown lists.

16

CSDS Worked Examples

 The atoms that are involved in the contact can be chosen by clicking on the atoms in the display window, but since in this case, the atoms chosen by default are correct, click on OK to proceed. Note, in this example we do not have the option of selecting a contact that involves hydrogen atoms. This is simply because hydrogens were not explicitly drawn when the carbamoyl group was defined; instead the hydrogens are viewed as a constraint on the nitrogen atom. Click on the Next > button to proceed to the final section of the Motif Search wizard. 4. Generating motifs We are interested in looking for motifs involving rings and infinite chains. Click on the check boxes for the ring motifs that contain 2, 3, 4, 5 and 6 contacts. Also click on the check boxes for infinite chains that repeat every 1, 2, 3, and 4 contacts.

CSDS Worked Examples

17

 Click on Next > to start the algorithm that takes the contacts and motif geometries requested and generates all the possible combinations of contacts that match the criteria. For example, checking the boxes as described above will generate 31 unique motifs. When the motifs have been generated, click on Next >. 5. Structure selection We now need to choose the structures that will be searched using this Motif query. We wish to search the entire Cambridge Structural Database, so from the Databases expandable list, click on the latest version of the CSD available and hit Select >>. Click Next > to continue. Finally, enter a name for the search, for example carbamoyl and begin the search by clicking on Start Search. 6. Viewing and analyzing the results. You can allow the search to run through the entire database (which may take a while) or you can stop the search early by hitting the Pause button at the bottom of the Searches tab within the Structure Navigator.

18

CSDS Worked Examples

 There are two ways of looking at the results of a search. Firstly we are going to use the View by Motif mode. Make sure by Motif is selected from the View results drop-down list towards the bottom of the Structure Navigator. View by Motif mode lists the motifs searched for, with the refcodes that exhibit the motif listed underneath. The selected structure, with the motif fragment of the molecule highlighted is shown in the display. The highlighting of the motif can be customised by clicking on the Options button at the top of the Structure Navigator and by selecting hit highlighting. Highlight the motif using the spacefill option. To view which motifs are formed most commonly, click on the % frequency column heading in the Structure Navigator. This should sort the motifs in ascending order of frequency of occurrence so the most common motifs are at the end of the list. We see that the most common motif is the R2 dimer motif, followed closely by the C1 infinite chain motif.

CSDS Worked Examples

19

 Both of these motifs have a reasonably high propensity to form, indicating that these motifs are potentially useful tools for engineering crystal structures. The least common motifs will include rings with 5 contacts. Since there is no crystallographic symmetry element with 5-fold symmetry, structures that display these motifs will commonly contain more than one molecule in the asymmetric unit. Check to see if this is so by clicking on the refcode names listed below the R5 motifs. If you switch on the colour by symmetry equivalence option, it is easier to see if symmetry independent molecules are involved in the motif. An example of a structure with the R5 motif, and which contains more than one molecule in the asymmetric unit is TAVKUX01.

20

CSDS Worked Examples

 To find the structures that contain both the R2 dimer motif and the C1 chain motif we can filter the search results. Click on the Options button towards the top of the Structure Navigator and select Filter Results. this will open the Filter search results dialogue. Drag and drop the R2 dimer motif and the C1 infinite chain motif to the Must have box.

CSDS Worked Examples

21

 Hit OK. The results of the filtering are presented in the second way of viewing structures, that of View by Structure mode. In the View by Structure mode the refcodes are listed and the motif(s) found within each structure are listed below the refcodes.

22

CSDS Worked Examples

 Click on the column heading number of motifs to sort the list in ascending order. Refcodes that contain two motifs will only contain the R2 motif and the C1 infinite chain motif. Click the Multi-view mode check box towards the bottom of the pane. This will allow motifs within a structure to be viewed simultaneously. Find a structure that contains only 2 motifs and expand the list to view the motifs. Click on each motif and you will be able to view how the motifs fit together in the crystal structure. If you look at the refcodes that contain more than 2 motifs you will notice that the R6(AB-ABAB-AB-AB-AB) motif is very common amongst structures that contain both the R2 motif and the C1 infinite chain motif. It is not unusual, if you search for more than one motif, to find that one motif belongs to a subset of another. If we filter the results of the search to remove all structures that contain the R2 dimer motif and the C1 infinite chain motif, we are left with a small percentage of structures. Looking through these structures it is interesting to note that the C2(AB-AB-AB-AB) motif is quite prevalent.
CSDS Worked Examples 23

 Thus we are beginning to build an understanding of how carbamoyl functional groups like to come together in crystal structures: dimers and infinite chains are very popular and these motifs dominate the packing. The difference between the C2 infinite chain motif and the C1 infinite chain is that two molecules, instead of one, define the repeat unit. This motif can occur when there is more than one carbamoyl group present on the molecule, or when there is more than one carbamoyl-bearing molecule in the asymmetric unit. However, it is these two simple motifs that dominate the packing: extended ring motifs often result from a combination of these basic units. This ends the tutorial.

24

CSDS Worked Examples

3.2 Tutorial 2: Studying Crystal Packing Features Introduction This tutorial studies a chlorophenyl stacking contact in the structure of 4-chlorobenzoic acid and how this compares to other halophenyl stacking contacts in the CSD. The structure is taken from Colapietro, M. & Domenicano A. (1982), Acta Cryst. B38, 1953-1957.

Menu Commands Required 1. Load the required crystal structure. Load the first structure for this tutorial by typing CLBZAP02 in the entry box at the top of the Structures pane within the Structure Navigator. This will show the structure of 4-chlorobenzoic acid in the Visualizer section of the display. 2. Study the packing contacts in the structure. Switch on the tick box next to H-Bond in the list box underneath the display area. This will display the hydrogen bonding interactions formed by the carboxylic acid group of the molecule.

CSDS Worked Examples

25

 Clicking with the left mouse button on one of the dashed red lines will show the molecule at the other end of the hydrogen bond - these interactions form a centrosymmetric hydrogen-bonded dimer between two molecules.

26

CSDS Worked Examples

 For the purposes of this tutorial we are interested in stacking contacts between phenyl groups, in order to display these interactions we need to generate a new specific contact. To generate a specific contact for the stacking interactions, click on the Contacts button in the Display Options pane, this will launch the Contacts dialogue box. Click on the Add button and choose Specific from the pull-down menu, this will display the Define Specific Contacts dialogue. Select to display contacts between C and C atoms with a distance range specified by Actual distance from 0.0 to 3.7 .

CSDS Worked Examples

27

 Next click on OK and then close the Contacts dialogue - the visualizer will now show stacking contacts perpendicular to the hydrogen-bonded ring. Expand the structure through these stacking contacts once in each direction and you will see that a 2D tape is formed by these intermolecular interactions.

28

CSDS Worked Examples

3. Perform Crystal Packing Feature search Turn off the display of contacts at this point by clicking on the checkboxes for H-bond and Specific Contact. Now select the six carbon atoms and one chlorine atom of the chlorophenyl group (atoms C1, C2, C3, C4, C5, C6 and Cl1) in each of two molecules related by a stacking contact.

CSDS Worked Examples

29

 These 14 atoms will now be used as the search parameters for similar packing features in the CSD. Choose Search from the top-level menu and select Crystal Packing Feature from the pulldown menu, this will launch the Packing Feature Search wizard. The first page of the dialogue will state that you have selected 14 atoms from 2 molecules from CLBZAP02, click Next > to proceed.

30

CSDS Worked Examples

 We will now relax the search criteria slightly to allow the chlorine atoms to be any halogen. To do this, select the two chlorine atoms in atoms pane (to select multiple atoms, left click whilst holding down the shift key) click on the Modify button and choose Element > More > Any Halogen from the resulting drop-down menus. Ensure that the Number of hydrogens and Number of bonds checkboxes are checked, then click on Next > to proceed.

CSDS Worked Examples

31

 Choose Very high similarity by clicking the relevant radio button under Level of Geometric Similarity Required and then click Next >. At this point you can choose one or more geometric parameters to record for each of the matched packing features that are found. Select the two chlorine atoms and click Add Distance > to choose this parameter, then click Next > to continue. Here the structures you wish to search for the defined packing feature must be selected. Add the full Cambridge Structural Database to the Selected Structures section by choosing the CSD from the list of available structures and clicking on the Select >> button. Finally, click Next > and then click on Start Search to run the packing feature search, hits will now start appearing in the Searches tab of the Structure Navigator. 4. Study list of search results Take a look at the list of hits shown in the Structure Navigator. There should be over 70 structures found with a similar packing feature in the database. Click on the RMS button at the top of the results list and the results will be ranked in order of the RMS deviation between the search packing feature and the hit packing feature. You will see that the halogen to halogen distance is also shown in the results list for each match. A glance at the structures with low RMS values (i.e. very similar packing features) shows that there are a number of other refcodes starting with CLBZAP, these are the same structure as CLBZAP02, but determined at different experimental conditions. Click on the Show packing feature checkbox at the bottom of the Structure Navigator pane and the search structure will be

32

CSDS Worked Examples

shown overlaid with the hit structure.

5. Identify closely related packing features Click on the button at the top of the refcode list to re-order the hits alphabetically by refcode and then select BRBZAP02, which should be listed as having an overlay RMS deviation of 0.06 . This crystal structure is of the closely related compound 4-bromobenzoic acid and it is clear to see that the two structures share a very similar halophenyl packing feature. Show the hydrogen bonds in the same way as before and expand the contacts in both structures - this will show that the structures also both contain the centrosymmetric hydrogen-bonded dimer. 6. Overlay crystal structures In order to investigate these structures further, it is necessary to expand out the network of molecules again. To make sure that the structures are easily distinguishable, click on the Options button in the Structure Navigator and select Hit Highlighting from the pull-down menu. Change the colour of the packing feature (4-chlorobenzoic acid) to green and then close the Hit
CSDS Worked Examples 33

Highlighting Style dialogue. The 2D tape common to these structures can be built up by expanding the contacts as shown before. Clicking on the tick box next to Short Contact in the display options will show that there is also a similar packing feature formed by CHO contacts between the phenyl and carboxylic acid groups. These stacking contacts, CHO interactions and hydrogen bonds form a double-layer of molecules all of which is conserved between the two structures.

Finally, add a further specific contact in the same way as before, except between atoms of type F, Cl, Br, I and with a distance range specified as 0.0 to 3.9 . Expanding these contacts will show that the double layers stack together in different ways between the two structures - the ClCl contacts are approximately linear, whereas the BrBr interactions have a Br-Br-C angle of around 90 . The structures therefore contain two entire conserved planes of interacting molecules, but the way that these stack is different. In CLBZAP02 the planes stack in an ABABABAB pattern, whereas in the BRBZAP02 structure the planes follow the stacking pattern ABCDABCD.
34 CSDS Worked Examples

 Show the Bravais-Friedel-Donnay-Harker morphology models for the two structures by clicking on Calculate in the top-level menu and selecting BFDH Morphology from the pull-down list. Interestingly the double layer orientations identified in the two structures also corresponds to the shape and orientation of the BFDH morphology (both are plates) as calculated by Mercury. BRBZAP02 is specified as having a crystal habit of plate, available under Structure Information by clicking on the More info button. Study of the paper for CLBZAP01 (Miller et al., 1974, J. Am. Chem. Soc., 96, 6334) shows that these crystals also grow as plates, with (100) being the developed face of the crystal (as suggested by BFDH). This ends the tutorial.

CSDS Worked Examples

35

3.3 Tutorial 3: Searching on Packing Similarity. Introduction Carbamazepine is a compound, which to date, has been found to have four polymorphs. Polymorphism, the ability of a molecule to crystallize in different three-dimensional structures is an interesting, though potentially troublesome phenomenon. Pharmaceutical companies are particularly interested in the polymorphic behaviour of target drug molecules. Inclusion of something other than the most thermodynamically stable polymorph in a tablet, for example, may lead to changes within the tablet over time, as the less stable polymorph converts to the more stable polymorph. Different polymorphs may exhibit different physical properties, such as the dissolution rate, which will affect the drug's efficacy. Menu Commands Required 1. Selecting structures for a packing similarity search Click on Search from the top-level menu and select the option Crystal Packing Similarity from the resulting drop-down list. This will open a dialogue where you can select structures upon which to perform packing similarity searches.

Click on Add Refcode in the Reference Structures pane and in the Enter Refcode dialogue type

36

CSDS Worked Examples

CBMZPN in the Refcode text box. We want to compare all carbamazepine structures with each other to determine which structures belong to which polymorph, so check the box Enter refcode family.

Click OK. A warning dialogue will be shown that tells you that the structure with the refcode CBMZPN does not have 3-D coordinates and therefore cannot be included in the search. Click OK to continue. Now we want to enter structures for comparison. As above, choose the full CBMZPN refcode family. 2. Starting the packing similarity search The default options for searching and comparing the structures are adequate for this search, so click on Compare to start the search. 3. Viewing and Analysing the results In the Structure Navigator, under the Searches tab, the results of the pairwise comparisons are listed.

CSDS Worked Examples

37

 The first comparison, that between CBMPZN02 and CBMZPN01, shows that these crystal structures are the same, with 15 out of 15 molecules overlaying with each other almost perfectly. The result of the comparison between CBMZPN03 and CBMZPN01, however, shows that the packing of only two molecules is similar between the structures, those involved in the amide dimer motif. Switch on H-bonds when viewing the CBMZPN03 and CBMZPN01 comparison to see the dimer motif.

38

CSDS Worked Examples

 It can be seen that for all the comparisons, the minimum number of molecules which the structures have in common is two. This is because the amide dimer motif is present in all of the polymorphic structures. To see which of the structures correspond to the different polymorphic structures, click on the Group by similarity check box towards the bottom of the Structure Navigator. The structures are now collected into 3 groups, suggesting that there are only 3 polymorphs.

CSDS Worked Examples

39

 This highlights an issue that can occur when comparing solvates, salts, hydrates or co-crystals containing just one component in common - large gaps can occur in the structures leading to slab-like, unrepresentative, clusters. In this case, CBMZPN03 has large voids in the structure due to disordered solvent molecules that were not modelled by the crystallographer, so it could alternatively be classed as a solvate. To perform a more accurate analysis, return to the search by clicking on the Edit Search button in the Post-Search Options pane. This time increase the cluster size to 20, run the analysis and group by similarity again, at which point the structures should now be grouped into 4 sets of structures. If we return to the list of pairwise comparisons (uncheck the Group by similarity check box) we can see that one of the comparisons, between structures CBMZPN03 and CBMZPN11, shows 19 out of 20 molecules in common. From the list of grouped structures we can see that these two structures belong to two different space groups, P-1 and R-3. In the list of pairwise comparisons, click on the entry with for structures CBMZPN03 and CBMZPN11. Switch on the display of H-bonds and switch packing on. To see how similar the structures are, click on Calculate from the top-level menu and select the option Packing/ Slicing. Increase the packing ranges to approximately 2.0 in the b-direction and -0.6 in the adirection. You can see that the molecules in the two structures continue to overlay quite well. If we increase the packing in the c or -c directions you can start to see where differences in the packing occur, with the molecules no longer overlaying well with each other.

40

CSDS Worked Examples

 How can it be that structures in P-1 and R-3 are so similar? First, reset the packing ranges to 0.0 to 1.0 on all axes by clicking the Reset button in the Packing and Slicing dialogue. Next, display the symmetry elements of the space groups by clicking on Display from the top-level menu and choose the option Symmetry Elements. If you change the colour of the 3-fold axis to pink, for example, the visualisation of the packing may be easier. Now switch on Colour by Symmetry Equivalence. You can see that the P-1 structure has four molecules in the asymmetric unit and these are coloured red, green, blue and yellow. It can be seen that three molecules of the P-1 structure, coloured red, yellow and blue overlay with the three molecules related by the three-fold screw axis of the R-3 structure. This can be seen particularly clearly if the packing ranges on the b and c axes are increased to 1.1 and 1.3, respectively.

CSDS Worked Examples

41

 Therefore, part of the reason the P-1 and R-3 structures are so similar is that the symmetry independent molecules of the P-1 structure display pseudosymmetry, approximating the 3-fold screw axis. As a further exercise, you can look for packing similarities between one structure of the carbamazepine family CMBZPN01 and the rest of the database. In this search, not only will you find matches to the other polymorphs, but also matches to hydrate and solvate structures. For example UNEYIV has 5 molecules packed similarly to CBMZPN01 even though UNEYIV is a dimethylsulfoxide solvate of carbamazepine. There are also some structures, for example TAZRAO, a cocrystal, where the amide dimer motif has been disrupted. However TAZRAO and CBMZPN01 do have a stacking interaction between two carbamazepine molecules in common. Can you find other examples where this motif is the common motif? This ends the tutorial.

42

CSDS Worked Examples

4. IsoStar Tutorials IsoStar is a library of intermolecular interactions. It is derived from data contained in the CSD and the PDB. It features: Intermolecular contact information for nearly 350 central groups and over 45 contact groups. Contact distributions displayed as scatterplots or density surfaces. Theoretical data for selected model systems. A simple web browser interface. 4.1 Tutorial 1: Using IsoStar Introduction In this tutorial you will learn how to: access a local copy of the IsoStar Home Page using a web browser (this should be in the browsers Bookmarks list). navigate the IsoStar pages, to locate and display scatterplots for contacts between isoxazole and various polar X-H groups in the CSD. manipulate the isoxazole scatterplots, using the IsoStar control window to display only the shortest contacts, identify and display the CSD entries in which these contacts are found and measure the contact distances. produce density surfaces from the isoxazole scatterplots and note the significant features. locate and display scatterplots for interactions between indole and N-H groups in the CSD and the PDB. Then manipulate these plots such that the underlying trends are clear and note the significant features. Menu Commands Required 1. Start your web browser and select the IsoStar Home Page from the Bookmarks list. 2. Select Ligand, Ring systems, N, O, C, H only from the list of central groups displayed on the left hand side of the IsoStar Home Page.

CSDS Worked Examples

43

3. A list of ligand ring systems containing N, O, C and H is now displayed in the main part of the browser window. Scroll down this list until you locate isoxazole and click on the link to display the isoxazole contact tables.

44

CSDS Worked Examples

4. Use the N-H hyperlink above the tables to move directly to the contact table for N-H and click on the link for any NH in the CSD column to display the scatterplot.

CSDS Worked Examples

45

5. The scatterplot is rotated using the left mouse button, translated using the middle (scroll) mouse button and zoomed using the right mouse button. 6. Click the vdW overlaps button in the control window to remove all contacts greater than the sum of van der Waals radii from the scatterplot. What does the scatterplot tell you about how N-H groups prefer to approach isoxazole? Is this what you expected? 7. Using the UpperLimit slider bar, remove all but the shortest one or two contacts from the scatterplot. Activate the Hyperlink checkbox above the scatterplot by clicking on it. This will switch on hyperlinking. Select one of the contact groups still displayed in the scatterplot using the mouse. The selected contact will turn green, and the contact, as observed in the CSD, will be displayed in the second visualiser window.

46

CSDS Worked Examples

 Note also that the tab in the Previous Files part of the IsoStar visualiser window has changed from Scatterplots to Hyperlinks, and that the CSD refcode is given in the CSD Identifier window. 8. A number of options that allow further control over the appearance of the contents of the visualiser window are available by right clicking anywhere in the visualiser. It is also possible to measure distances, angles and torsions via this menu. The contact distance in the hyperlink window can be measured in the following way: Ensure the Hyperlink window is the active one either by clicking in the window itself or by hitting the Current visualiser radio button. Right click in the window, and select Measure then Measure Distances from the pull-down menu. Now determine the length of the N...H contact distance by clicking on both atoms that are involved in the contact. Close the visualiser window and return to the IsoStar browser window. 9. Display the scatterplot for contacts between isoxazole and any polar X-H in the CSD (this contact group is in the General contact table). Select the Contours button in the Scatterplots tab. This will open a Contour Surfaces window where it is possible to add contours to the scatterplot and configure them. Ensure the Internal Scaling radio button is activated, select a colour (e.g. blue) using the Color button, set the scaling level at 20 then hit Create.

Blue contours will appear on the scatterplot. Level 20 implies that it is 20 times more likely than random chance that a contact will occur within the blue contour.

CSDS Worked Examples

47

 In a similar way, introduce a red contour at level 40 and a yellow contour at level 80. Now hide the scatterplot by clicking on the Hide All tick box in the main IsoStar control window.

How does this plot compare with the plot for N-H donors only? Is this what you expected?

48

CSDS Worked Examples

10. Close the IsoStar control window and return to the browser. 11. Find the central group indole (see below; Ring systems, N,C,H only).

12. Display scatterplots for the interactions between indole and any NH in the CSD and in the PDB. Compare the two plots. What are the differences? 13. Produce a density surface from the indole...any NH CSD scatterplot (click on the Contour button and use the same contour settings as before, i.e. blue: level 20, red: level 40 and yellow: level 80). The blue contours can be hidden if required by de-activating the tickbox under the Show column in the Contour Surfaces window. What does the density surface tell you about interactions between indole N-H? Is this what you expected? This ends the tutorial.

CSDS Worked Examples

49