Hi,
I currently have a severe refinement problem. My crystals give high resolution data and good merging statistics but however, the R-free gets stuck at around 28-35%.
Details:
Resolution of the crystals varies between 2.1 and 1.4 Angstroem.
The unit cell contants are the following: a=55 b=55 c=229 in sg P4(3)2(1)2.
Data collection:
Concerning data collection the longer c-axis has been taken into account, especially for the low resolution!
Typical datacollection parameters at the SLS PX1 beamline (MARCCD-detector) are:
Dphi=0.25°, dist=200mm (for an image see marccd_slspx1.jpg).
Inhouse datasets (with MARDTB) quite often seem to be better concerning the final R-free values (Dphi=0.25°, dist 175-200mm). For processing of the data the XDS package is used (xds_process.txt).
Structure solution:
The structure was solved by molecular replacement (Molrep) in sg P4(3)2(1)2. The CC was 58% (best wrong 16%) and the R-factor was 41% (best wrong 59%). The asymmetric units contains one molecule.
Refinement:
Refinement goes smoothly until it gets stuck at around 30% (see refinement_p43212.txt), which is in my eyes too high for these resolutions. The twenty N-terminal residues (of 330 amino acids in total) are only badly defined.
The general quality of the density improves when going to higher resolution (R-free is then often around 35%). Technically following refinement packages have been used: CNX, Refmac and Shelx. In Refmac 0.05-0.1 has been used as weighting matrix and gives bond lengths rmsd's ~0.016 and bond angle rmsd's ~1.5.
Refinement has also been tried in sg P4(3), C222(1) and P2(1)2(1)2(1) and gets stuck at 30% R-free. Sg C222(1) has been tested, as this is the sg that xprep suggests (see xscale.prp).
Twinning:
The possibility of merohedral twinning was also taken into account, but does not seem to be the case (see twinning_cnx.txt, twinning_xprep_p43.txt). The 2nd moment of I for example has values around 2.1 (ccp4 truncate in sg P4(3)2(1)2)). For testing the twin refinement, sg P4(3) was used with two molecules per asu (CNX: <|I|^2>/(<|I|>)^2 = 2.2541; SHELX: BASF value = 0.428 [C] or 0.408 [NC], <|E^2-1|> 0.751/0.749.
Crystal morphology:
The morphology of the crystals seems to be quite unique: an UFO-shape with scales (see xtal.jpg).
When taking a look at the diffraction image (marccd_slspx1.jpg) one gets the impression of a diamond/rhombus shaped background underlying the spots (streaks). Could this be a problem? If, I would expect higher Rsym's?
What else could be the problem? Did we oversee anything? Any suggestions are welcome!
Best wishes,
Dennis Fiegen
P.S.: Due to size restrictions of this email I did not attach the mentioned files, but anyone who is interested in more detailed information can request them.
Dr. Dennis Fiegen
Protein Crystallography/Structural Research
Boehringer Ingelheim Pharma GmbH & Co. KG Deutschland
Birkendorferstrasse 65
88400 BIBERACH, Germany
tel. +49 - (0)7351 - 54 98634
fax. +49 - (0)7351 - 83 4683
email [EMAIL PROTECTED]
