[sdpd] DIRDIF applied to SDPDRR sample 2

[Armel Le Bail <armel... @fluo.univ-lemans.fr>]

Dear Members,

Already 57 subscribers to the SDPD Mailing List, 36 hours
after its birth. This may be considered as quite encouraging !

Unfortunately for you, descriptions of efforts in trying to
succeed in a SDPD may be really boring. I hope that you are 
prepared to read long emails ! Here is the first one :-).

Since the SDPD Round Robin deadline, one of my constant
preoccupation is to try to demonstrate that the two
structures were not so difficult to determine from the
provided data (which were not presenting the highest
resolution possible, otherwise there would have been
no challenge at all).

About the pharmaceutical sample, tetracycline hydrochloride,
you may remember that 2 participants of the Round Robin
were able to solve it. One by the Global Optimization
Method for molecule location (DRUID software), and the 
other by Patterson method, using the CSD package. The
SDPD Round Robin organizers were also able to solve
the structure by using PATSEE.
For the SDPDRR see :
    http://www.cristal.org/SDPDRR/

>From my quest in order to find available software able
to solve this problem easily, I have one to present today :
DIRDIF. It is a freeware, distributed with source code
at :    http://www-sci.sci.kun.nl/xtal/dirdif/dirdif.html

DIRDIF, from Prof. Paul T. Beurskens and co-workers is a 
computer  program system for crystal structure determination by 
Patterson  methods and direct methods applied to difference 
structure factors.  Versions are available for a wide range of 
computer and  operating systems. They have been tested and 
used on IBM/cms,  VAX/VMS, CYBER computer, PC (PCDOS, 
MSDOS) and various unix systems (Silicon Graphics, SunOS, 
Solaris, DEC, Alpha, HP, linux, IBM/AIX).

A MS Windows version, built by Louis Farrugia, is available
at :         http://www.chem.gla.ac.uk/~louis/software/

None of the 31 participants (of 72) having answered to the
questions concerning how they intended to solve the SDPD
Round Robin sample structures suggested DIRDIF ! This
means that DIRDIF is not well known by our community,
and indeed I ignored myself what could be its capacities
till last week...

Owing to the poor resolution data, it is clear that methods
allowing to use knowledge of the molecular structure have
an advantage here. This is the case of DIRDIF presenting
an option for structures of molecules with (partly) known
geometry.

Once identified that the C22H24N2O8.HCl was undoubtedly
built from the very well known tetracycline molecule (looking
inside CSD or interrogating Internet, or chemical company
catalogs, for instance), all the DIRDIF user has to do is to
insert the model inside a file gathering molecular fragments,
if it is not already included (ORBASE and ORUSER files).

For the SDPDRR sample 2, I used the tetracycline molecule
as provided in the TETCYH10 CSD entry, which is the
hexahydrate (the same was used by the DRUID team). The
data in the .mol2 file built at the end of  a CSD query were
inserted in the DIRDIF ORUSER file, excluding the water
molecules, of course. Then, I used the DOS DIRDIF version
(quite easy to install), after having prepared a INS and a
HKL files (both SHELX -like, the latter from a Le Bail
method structure factor extraction from the synchrotron
data, with application of the OVERLAP software in order
to exclude reflections having neighbouring ones at less
than 0.005 2-theta degrees - yes, no error - synchrotron
data ;-).

I really found the DIRDIF manual clear, there is an
example provided, and typical runs are described. For
obtaining the molecule location, the sequence of orders
at the command line are (neglecting some questions/
answers):

DIRDIF TETRAC PATTY
This produces Patterson search. Examining the result shows
that the Cl atom is located, but the further efforts of DIRDIF
in order to propose a model are not very convincing. Only 2
files are called here (INS and HKL). TETRAC is an
arbitrary chosen name.

DIRDIF TETRAC ORBASE
This will prepare data including the tetracycline fragment
entered previously in ORUSER.

DIRDIF TETRAC ORIENT
This will execute the search for molecule location/orientation

One of the location proposition is used by DIRDIF for
trying to find the lacking atoms. DIRDIF successfully
locates the Cl atom. However, you may have to disregard
the final rearrangements of atoms proposed after several
Fourier syntheses, and preferably look at the coordinates
of the molecule as oriented (and not further distorted).

A Zipped file containing the three files (INS, HKL and 
ORUSER) which will allow you to reproduce my efforts is
available at : 
    http://www.cristal.org/sdpd/data/dirdif-1.zip
The file contains also atold (summary of DIRDIF propositions)
as well as dirdif.pcr and dirdif.dat which would
allow you to verify that the raw model provided by DIRDIF
leads to Rp =32%, RF = 16.4%, by using FULLPROF on the 
SDPD Round Robin synchrotron data, a quite good starting 
model for subsequent refinements, no ?

Some words about the Louis Farrugia Windows version.
There is a bug making that the ORIENT version is not
working. Louis is currently busy on that question. Wait
before downloading. But the DOS version works quite
fine.

All the best, and have fun with DIRDIF.

Thanks to the DIRDIF team for making available the
whole package in the old good Academic way (including
sources, so that the curious may understand more) !


Armel Le Bail - Universite du Maine, Laboratoire des Fluorures,
CNRS ESA 6010, Av. O. Messiaen, 72085 Le Mans Cedex 9, France
http://www.cristal.org/