DSR: enhanced modelling and refinement of disordered structures with SHELXL

The new computer program DSR enables semi-automatic modelling of disordered or well ordered moieties in crystal structures via a placement procedure of molecular fragments and corresponding stereochemical restraints from a database.


[atom n]
Minimum three atoms each (including Q-peaks). Source and target atoms have to include the same number of atoms or Q-peaks. Target atoms can be either regular atoms or atoms in residues. Atoms in residues can be addressed by the "_" notation. C1_2 would be atom C1 in residue number 2.

PART n
Optional SHELXL PART definition.

OCC mn
Optional occupancy and free variable definition for the fragment.

DFIX
Optional, generates DFIX/DANG restraints instead of those from the database. All 1,2-and 1,3-distances in the fragment and their neighboring atoms are restrained with DFIX and DANG respectively. Bonds to symmetry equivalent atoms are ignored.

RESI class num [alias]
Optional residue definition as in SHELXL.

Use of Residues
To use the RESI command in DSR has several advantages. It places the fragment into a residue and therefore no renaming of the atoms in the fragment is performed by DSR. If residues are used, the restraints like "SADI_class Atoms" are inserted only once, since they act on the atoms in all residues with the same class. Residues are especially useful, if the same moiety is repeated several times in a crystal structure. Different moieties of the same residue class are distinguished by different residue numbers. Residue number must be unique in a .res file. The DSR command RESI without any further options is normally the best practice. DSR then uses the residue class name from the database and assigns the lowest unused residue number automatically. But the user can also specify a particular residue class and/or number after the RESI command, if desired.
The RESI option can be used in three ways: 1) If only a RESI command is given (best practice), the residue class is taken from the database entry and the residue number is automatically generated. 2) If RESI with only a number is given, DSR takes the residue class from the database with the given number.
3) RESI with a number and a class overwrites the information from the database and gives complete control over the residue.

Residues in structure refinement
The concept of residues is well established in structure refinements of macromolecules. Although it simplifies bookkeeping and restraining of complex structures, it is rarely employed by chemical crystallographers, who are mostly dealing with small molecule structures. For using the RESI command of DSR, the user should have at least basic knowledge about the concept of residues in SHELXL. The manual on the SHELX website provides more detailed information: http://shelx.uni-ac.gwdg.de/SHELX/wikis.php Employing identical atomic labelling schemes for several different parts and residues of a structure in a SHELXL refinement only requires a single restraint dictionary to restrain all occasions of that residue class in the same way, which significantly supports handling of large and complex structures in a consistent manner. As small molecule programs written ages ago were initially not designed for the residue concept it should be emphasized that the most elegant way for producing a crystallographic information file (CIF) suitable for other programs (e.g. CHECKCIF) from such a refinement is the "ACTA TABS" command (applies only to SHELXL versions 2014 or later). Additionally, disordered residues should be numbered in between 1 to 9. Atoms names in the resulting CIF can otherwise contain more characters than expected by some programs. For more details please see the SHELXL instructions at http://shelx.uniac.gwdg.de/SHELX/shelxl_html.php#ACTA.

Further Details to the OC(CF3)3 Fragment Placement
Disorder of a perfluorinated tert-butyl group is frequently observed in the [Al(OR F )4]type of weakly coordinating counterions (Krossing, 2001;Bihlmeier et al., 2004;Köchner et al., 2012). CCDC 936462 was first published in (Lichtenthaler et al., 2013) and contains the full crystallographic data of the structure chosen to demonstrate placement of the second OC(CF3)3 fragment on the second position using DSR. The following command line would be needed in the SHELX .res file: REM DSR PUT OC(CF3)3 WITH O1 C1 C2 C3 ON O1_1 C1_1 Q4 Q7 Q6 PART 2 OCC -21 RESI A disordered fragment requires a free variable for the site occupancy and a part number, which is accomplished by the OCC and PART commands. For example OCC 21 sets the free variable number 2 for the fragment and the occupancy to one. Non-existing free variables are introduced to the FVAR instruction automatically.
DSR inserts the following two entries into the res file. The FRAG ... FEND block defines the transferred fragment through the cell parameters and atom coordinates. The other text block beginning with RESI defines the residue, the part, the free variable for the site occupancy and the position of the fitted fragment. The complete fitting process takes less than a second, even on slow computers. For smaller molecules in lower occupancy or diffuse environment, the use of DFIX bond restraints might be useful. Therefore, the DSR command DFIX would ignore all restraints in the database and generate DFIX/DANG/FLAT restraints for all 1,2-and 1,3-bond distances occurring in the fragment and its neighboring atoms.

Further Details on the Supramolecular Clathrochelate Fragment (PM41) Placement
In this structure 64% of main residue disorder is triggered by the disorder of Cd ions over a special position (inversion centre) to which the supramolecular Clathrochelate ligands are coordinated. CCDC 980636 contains the full crystallographic data of this structure, which was first published as MOF 14 in (Pascu et al., 2014). To place the second PM41 fragment on the second position with DSR, the following DSR command line would be needed in the SHELX .res file: REM DSR PUT PM41 WITH N4 B7 N31 B20 ON Q7 Q133 Q44 Q171 PART 2 OCC 10.5 =

RESI PM4
In this case of disordered near a special position the occupancy of each of the two parts must be 0.5 (50%), which is accomplished by the OCC 10.5 command. After fragment placement (FRAG ... FEND) and lifting of constraints from the initial body refinement (AFIX), DSR also writes the following list of stereochemical restraints from the DSR database. As the list contains 206 lines for such a large fragment, all restraints can be written into a text file in the same folder named "dsr_PM41.dfix", which is automatically included in the SHELX .res file by DSR via the "+dsr_PM41.dfix" instruction. "_PM41" is a reference to all residues named "PM4", independent of residue number or PART number. Using the identical atomic labelling scheme in all residues and parts of all PM4 moieties ensures that only one restraint dictionary is required for all of them. (see above +section "Residues in structure refinement") Restraints in the "dsr_PM41.dfix" file:

Database
Three typical examples of database entries are listed below. The lines starting with "rem Src:" and "rem Name:" are not required but provide additional information for each database entry in the supplied main database. The name in the "rem Name:" line also appears in the listing of the "dsr -l" output and can be searched with the -s option. 1 0.880000 -0.330900 0.267190 1.000000 0.05000 C2 1 0.786200 -0.377700 0.238080 1.000000 0.05000 C3 1 0.760600 -0.318400 0.192570 1.000000 0.05000 C4 1 0.829200 -0.212200 0.175520 1.000000 0.05000 C5 1 0.923200 -0.164400 0.204000 1.000000 0.05000 C6 1 0.948800 -0.223200 0.249920 1.000000 0.05000 </Benzene> Fragment Import from GRADE Web Server DSR can import molecular fragments directly from the .tgz file generated by Grade Web Server of Global Phasing Ltd. (Smart & Womack, 2014) from which it reads the .dfix file for the restraints and runtime information, the .pdb file for the coordinates as well as comments and the name. All these information are then stored in the user database "dsr_user_db.txt". As it is essential that the fragment has proper names before being imported to the DSR user database, appropriate compound and residue names should already be provided to the Grade Web Server. Additionally, the output of SHELX restraints (e.g. non-hydrogen atoms only) has to be activated.

Fragment Export
The option -c exports the fragment to the clip board. This clipboard entry can then be used for the match functionality in the program Olex2 (Guzei, 2014;Dolomanov et al., 2009

Automatic Restraints
For molecules with low occupancy or in a diffuse environment, the use of explicit distance restraints (DFIX, DANG) can be more powerful than similarity restraints (SADI). Therefore, the DSR command DFIX ignores all restraints in the database and generates DFIX/DANG/FLAT restraints for all 1,2-and 1,3-bond distances and planar bonding environments occurring in the fragment and it's neighboring atoms. This implies that the bond distances of the database fragment should be of sufficient accuracy to result in correct bond restraints. Especially the bond distances to other moieties are only as accurate as the initial fragment fit. Therefore, DSR warns about a possible inaccurate restraint in this case. Nevertheless, the main purpose of the automatic restraints is to get a stable starting model which needs to be further optimized in the subsequent refinement. DSR generates the DFIX/DANG restraints automatically during the fragment fit by analyzing the fitted fragment and its surrounding atoms in the connectivity table for their next and second-next neighbors with a combination of the connectivity list from the SHELXL .lst file (1,2-distances) and by graph theoretical concepts via the NetworkX software library (Hagberg et al., 2013). An adjacency matrix of the connected atoms is generated with the atomic distances as weight. This matrix allows to obtain the 1,3-distances and finds rings, which are inside the fitted fragment and its neighboring atoms. The planarity of rings is analyzed similar to the FLAT restraint in SHELX by the volume of the tetrahedron spanned by groups of four atoms from the ring. FLAT restraints are generated for them accordingly. For example, the distance restraints generated for the example molecule of the main paper are as follows: The distance restraints from the fitted fragment to the main molecule e.g. DFIX 1.6857 O1_4 AL1_0 should be carefully revised. They are only as accurate as the fit of the fragment.

Naming Scheme
Without using residues during a fragment fit with DSR the program defines a unique set of atom names to avoid clashes with the already existing atom names. Therefore, DSR numbers each element from 1 to the number of atoms of this element, than compares the names with the existing atom names. If an atom name already exists, a character from A to Z is appended to the name until every atom name is unique.