SYSTEM-S

Look here for an alternative System-S tutorial by Lachlan Cranswick (CCP14).

========

1 - WHAT IS 'SYSTEM-S'

***** S-USERS SHOULD BE REGISTRATED SHELX-USERS ! *****

The external tools (SHELXS97, SHELXL97, SHELXD, DIRDIF99, SIR97, SIR2002) are used 'as supplied' by their respective authors and should be obtained from the sources given below. They are NOT provided as part of this distribution. S will NOT work when no SHELX-97 executables are found in the PATH.

* WARNING: S IS STILL UNDER FURTHER DEVELOPMENT *

SYSTEM-S (or S for short) is a closely SHELX(S/L) compatible atomic resolution single crystal structure determination suite providing a unique crystallography aware SHELL functionality for the invocation and management of the 'tools-for-the-job'.

SYSTEM-S is for the professional chemical crystallographer who is bored by the routine input/output preparation/analysis cycle, but knowledgeable about the procedural details and aware of all signals of possible trouble.

SYSTEM-S operates as a supervisor (with a crystallographic degree and with an organized memory) over the process of a single crystal structure determination by X-ray diffraction techniques.

SYSTEM-S provides an interface to a variety of established datareduction, structure determination, refinement and evaluation tools.

SYSTEM-S provides an environment in which it is made easy to tackle 'hard'-structures requiring multiple attempts to solve a structure in different space groups and with different structure solution methods.

SYSTEM-S provides three modes of operation:

Guided Mode: The program suggests the course of the structure determination on the basis of the current context.
Automatic Mode: An automatic structure determination from primary data to final (refined) results is attempted. This mode will be referred to as the No-Questions-Asked (or NQA) mode.
Silent Mode: Automatic 'filter' mode run from the command line.

The NQA-mode (and Silent mode) obviously works only in the case of (relatively) trouble free structures (i.e. no disorder, twinning and similar specialist issues).

The tools available within SYSTEM S are either build-in (i.e. part of PLATON) or external (i.e. plug-ins).
External:

SHELXS97 - http://shelx.uni-ac.gwdg.de/SHELX/
SHELXL97 - http://shelx.uni-ac.gwdg.de/SHELX/
SHELXD - http://shelx.uni-ac.gwdg.de/SHELX/
SIR97 - http://www.ba.cnr.it/IRMEC/SirWare.html
SIR2002 - http://www.ba.cnr.it/IRMEC/SirWare.html
DIRDIF99 - http://www-sci.sci.ru.nl/xtal/dirdif/

Build-in:

PLATON - Geometry Calculations etc.
ORTEP - Stripped ORTEP
SHXT86 - Stripped, Adapted & Modified old but powerful SHELXS86/TREF.
PLUTON - Molecular Graphics
HELENA - CAD4-datareduction
EXOR - Work-up/completion of raw E-map's
ABSTOMPA - Exact correction for absorption
DELABS - Empirical correction for absorption
Both SYSTEM-S commands such as SHELXL ISO 3 (for three cycles of isotropic least squares refinement with the program SHELXL97) and standard UNIX-SHELL commands such as ps -aux (to see what is going on) can be given on the S-SHELL prompt.
SYSTEM-S sets up a directory tree for the storage of information supplied and generated during the structure determination and manages input/output for the available set of crystallographic tools.
SYSTEM-S restarts, when re-invoked with a compound name in the context where it was terminated during the last invocation. The current context can be reset to an earlier context with the RELINK, TRMX and SPGR commands.
SYSTEM-S is the default tool used and developed in the context of our National Service Crystallography Center (Utrecht, The Netherlands) handling over 200 (mainly organometallic) structures per year and building on experience gained from over 2000 structure determinations.
SYSTEM-S comes as a build-in feature of PLATON.
SYSTEM-S is (currently) available only on the UNIX/LINUX Platform due to extensive use of typical UNIX-SHELL features (e.g. sort, ln).

2 - IMPLEMENTATION REQUIREMENTS

It is assumed by default that all programs called from PLATON/S are in the path so that PLATON/S can find them.

If not, environment parameters should be set:

platon or PLAEXE for PLATON (Multipurpose Crystallographic Tool)
shelxs or SHSEXE for SHELXS97 (Direct Methods)
shelxl or SHLEXE for SHELXL97 (Least-squares Refinement)
sir97 or SIREXE for SIR97 (Direct Methods)
sir2002 or SR2EXE for SIR2002 (Direct Methods)
dirdif or DIREXE for DIRDIF96/99 (Heavy Atom / Patterson Methods (Orient))

Example for setting PLAEXE in the (t)csh-shell:

setenv PLAEXE /mnt/spea/bin/platon

when the platon executable is in /mnt/spea/bin

3 - DEMO run(s) of S for C(10)H(12)O(6)

Assumptions:

s is in the PATH (e.g. with the soft link to the platon executable i.e. 'ln -s platon s')

shelxs and shelxl are in the PATH (or set with the environment variables SHSEXE and SHLEXE respectively).

The two files sdemo.ins and sdemo.hkl are available in the current working directory from which S will be invoked.

sdemo.ins contains relevant information for a small organic compound (with formula C10H12O6), on the cell, the wavelength ,info on the expected cell-content and HKLF 3 or 4 type, all in SHELX-style.

TITL SDEMO
CELL 0.71073 4.0007 7.7300 16.7597 90.0540 94.0760 90.0528
ZERR 2 0.0010 0.0008 0.0022 0.0096 0.0160 0.0144
SFAC C H O
UNIT 10 12 6
HKLF 4

Note: no spacegroup information required. Data/instruction lines other than the above are ignored. The CELL should be consistent with that of the reflection file.

sdemo.hkl contains a standard shelx HKLF 4 style dataset

S can now be run in the auto-mode (the No-Questions-Asked mode) with the keyboard instruction:

s sdemo.ins nqa

After some time (during which the space group is determined, the structure is solved by direct methods (SHELXS) and refined (SHELXL) including H-atoms) the result of the analysis is shown as a rotating molecule.

The rotation can be stopped by clicking in the window.

S can be terminated by typing 'END'.

Alternatively, S can be started in the 'guided' mode via:

s sdemo.ins

An interactive sequence is set up in response.

User input routinely involves 'hitting-the-return-key' when the suggested material within [] is o.k (or clicking on ACCEPT-DEFAULT in the menu bar).

An 'END' instruction terminates S.
S will return to the status where it was when restarted.

Remarks:

All calculations are done in a subdirectory of the directory 's'.
Directory 's' will be created automatically in the current directory unless there exists already such a directory in '~USER'.
Calculations on the sdemo example may be restarted in the context where it was left with
's sdemo'.
To remove the 'sdemo-stuff' from this tree goto directory 's' and do a
'rm -r sdemo'
PLEASE NOTE: S is not finished, there are several loose ends where more work has to be done or in progress (non-standard refinement, absorption correction other than DELABS etc).
Final refinement can be taken up with the files in 's/COMPOUND/tm/sg/shelxl'.

A second example (C13 H24 N2 Pd), illustrating a structure determined automatically by heavy atom methods (DIRDIF99), can be run with sdemo1.ins and sdemo1.hkl.

4 - Worked Example in the Guided Mode

Start the structure determination with s sdemo.ins

The s-shell prompt should look like s[CELL] (Fig.1)

Hitting the return key (or clicking on ACCEPT-DEF) will bring up the section of S that establishes the cell dimensions and associated esd's. In the current case we just acknowledge (with a return) the correctness of the values found in sdemo.ins. When desired, cell data may be changed at this point. (Fig.1a)

The next suggested logical step, i.e s[TRMX], (Fig.2) is the determination of the lattice type and Laue group with associated transformation matrix. Hitting the return will bring up a number of options. (Fig.3) The suggested choise (#1) is accepted again with a RETURN. The second option (#2) could be attempted lateron when #1 doesn't lead to results. (Fig.3a)

The next step s[SPGR] brings up the section of the space group specification. A number of choises, (Fig.4) based on the observed systematic extinctions, is suggested, along with an a-priori choise (#14).

The next step s[FORMULA] brings up the section to specify the cell content. In our case, the formula given in sdemo.ins is suggested. (Fig.5) However any other specification is possible. Here we accept the suggestion.

This brings up [Z]. Hitting RETURN will generate a suggested value for Z (i.e. 2). Any other reasonable value may be entered here. In our case we just hit RETURN again.

This brings us to the core of a structure determination, i.e. the phase determination. S suggests to run SHELXS for this. (Fig.6) Alternatively, the older SHELXS86 (as opposed to SHELXS97), SIR or SIR2002 could be attempted when available on the machine. Here we again take the default choise.

The result of the SHELXS/TREF calculation is now shown for inspection with PLUTON. (Fig.7) PLUTON can be terminated by clicking on EXIT.

The list of atoms generated with SHELXS obviously needs some 'cleanup'. This can be done with a procedure called EXOR (short for exorcise).

Indeed, all noise peaks have been removed and element types correctly assigned as shown in (Fig.8) If not, as may be the case with more problematic structures, PLUTON may be used to RENAME atoms to their desired labels and atom types for those mis-assigned by the automatic procedure. Also, remaining ghosts may be removed from the current model (stored in s.res; and optionally to be inspected by clicking on the LstRES Menu-button) with PLUTON.

On the termination of PLUTON, S suggests isotropic refinement as the next step:
s[SHELXL ISO].

At the end of the refinement a difference map is calculated from which the highest peaks can be appended to the parameter file to be inspected with PLUTON. In the current case, no significant residual density is found. Hit RETURN.

This step is followed by anisotropic refinement indicated with s[SHELXL ANISO].

In the next step it is suggested to find H-atom positions in a difference fourier map which is effected by hitting the RETURN key on s[HATOMS]. The result is again shown in a PLUTON plot. (Fig.9) The atom list may be edited in this stage.

Intermediate SHELXL refinement results and warning display look like (Fig.9a)

In the next step, H-atoms are included in the refinement.

The final step involves a weighted refinement. (Fig.10) This step can be repeated until complete convergence.

Terminate with END.

The results of the current refinement are in ~USER/s/sdemo/tm/sg/shelxl.

5 - DIRECTORY STRUCTURE of SYSTEM S

S operates in a directory structure under 'CURRENT_DIR/s/' (or '~USER/s/').
Note: '~USER/s/' is always used once this directory exists (either deliberatery or as a result of a user error)

Each project (structure) has its own subdirectory tree starting with the name of the structure (e.g. sdemo). The top-level directory for a structure is indicated as 'level-0'

Sub-directories of a level-0 directory include level-1 trees (one for each lattice type that is attempted).

Sub-directories of a level-1 directory include level-2 trees (one for each space group attempted to solve the structure in).

Level-2 houses subdirectories for the varies structure determination and refinement tools and there associated data.

The TREE instruction provides a display of the tree structure.
(Note: TREE does not display hidden files (stating with .) that should not be touched, since they perform 'memory'-functions)

The complete directory tree for a given compound compound can be removed either from buttons on the SYSTEM-S menu or via the command line instruction:

s compound remove

6 - PRIMARY (RAW) DATA

S provides three options for the storage of the original (raw) diffractometer data from which S can be started.

With COMPOUND.ins and COMPOUND.hkl (or COMPOUND.fcf) data in the current directory
Example: s sdemo.ins
With the raw shelx reflection data stored in a directory below the default (i.e. /mnt/shxdata) location.
E.g. Shelx data (i.e. shelx.hkl and shelx.ins) for compound s1000 are stored in /mnt/shxdata/s1000
S will find the data (shelx.ins & shelx.hkl) when started up with s s1000
The default location can be changed by setting the environment variable SHXPATH to the proper alternative.
In our local operation, CAD4 files (e.g. s1000.cad) are stored in a single directory /mnt/cad4data (i.e. /mnt/cad4data/s1000.cad).
System S is started in this context as s s1000
An alternative is to create an 's1000' sub-directory under 's' and copy the CAD4 data as s1000.cad into '~s/s1000'.
System S is started as s s1000.

7 - OTHER TRY-IT-YOURSELF EXAMPLES

Following are adapted (i.e. no prior space-group information given ins the '.ins' data-file) datasets supplied with Dr A.D. Hunter's SHELXTL course.

Example1: sfun1.ins and sfun1.hkl

Example2: sfun2.ins and sfun2.hkl

Example3: sfun3.ins and sfun3.hkl

Example of an inorganic compound Cs2TiSi6O15: csti.ins and csti.hkl

8 - How to Implement the new Space group suggested by ADDSYM

ADDSYM (either run explicitly or implicitly) will suggest an alternative spacegroup in RED on the main S window. Addsym is invoked automatically (unless switched off) at the start of the anisotropic refinement.

The transfer of the current structural results to the suggested space group is effected by clicking on 'TRMX' and 'SPGR'.

The Formula and Z can be adapted when desired following this transfer.

Refinement can now proceed in the new spacegroup.

Example: Solve the 'sdemo' structure not in P21/c but in Pc (# 7). At the anisotropic refinement stage, a message 'M/P P21/c' in RED will appear to attract attention to the possibly missed or pseudo-symmetry.

9 - How to run S on a 'CIF/FCF' Dataset

System S can be invoked using a 'fcf' (and 'cif') formatted file(s) (e.g. taken from the IUCR-Acta Cryst C Web-page) via:

s demo.fcf

S will convert the data in the .fcf file into a shelx.hkl (HKLF 4 format) file and ask for additional missing data.

The .hkl file can be found in the subdirectory 'hklf'.

A shelx.ins file is prepared from the data available in the 'fcf' file. Since an 'fcf' file doesn't contain information on the composition composition, this info should be given manually.

Alternatively, when both a 'CIF' and an 'FCF' (either with extension '.hkl' or '.fcf' is available, S can we invoked with:

s demo.cif

10 - KappaCCD/Denzo to PLATON/SYSTEM-S interface

The KappaCCD/Denzo software provides two output file formats

compound.hkl - i.e. SHELXL structured file. In order to run SYSTEM-S, an additional compound.ins has to be prepared.
import.cif - i.e. a reflection CIF containing both reflection data, cell dimensions, wavelength and optionally the supposed composition formula.
SYSTEM-S can be started for this dataset via platon -s import.cif.

11 - SUMMARY OF S-INSTRUCTIONS

Warning: Not all options available yet in this version !!!

ABSGAUSS

Absorption correction using a Gaussian Integration Grid.

ABSTOMPA

Absorption correction following the de Meulenaer & Tompa analytical correction technique.

ASYM (VIEW)

Extensive listing and display of the averaging and completeness of the reflection data as supplied.

The VIEW keyword invokes a display function giving a detailed view on the reciprocal lattice (completeness etc.)

AUTO

Synonymous for the No-Questions-Asked mode.

DELABS

Empirical absorption correction following the modified DIFABS (Walker & Stuart) approach.

DELABS is suggested by S in the isotropic refinement stage (and can be ignored when desired by asking for 'SHELXL ISO' as the next calculation.

DELABS can be repeated in any subsequent refinement stage. It always starts from the primary reflection file as supplied (with direction cosines). Anisotropic thermal parameters are automatically converted into corresponding isotropic U(eq) values before the DIFABS calculation.

It is a good idea to repeat the DELABS procedure when all scattering power is accounted for (including H-atoms).

DIRDIF

The method of choise for heavy atom structures.

DIRDIF may have problems with structure determinations run with an incorrect CONTENTS formula, in particular when the number of heavy atoms is different from the number suggested.

EXOR

Work-up of the 'raw' peaklist of a structure determination package (e.g. SHELXS).

Atom types are assigned to the resulting peaklist on the basis of the contents formula.

The correct identification of a peak as C,O or N may be hampered by difference in thermal parameters (i.e. periferal O atoms may fit the peak height of a carbon atom and a central C may fit the peak height of an O atom.