Dear Eleanor,
no I was not talking about a cubic SG (in my case, as an example, it's actually orthorhombic).
I think that the most likely explanation is that there was indeed a bug, at least on Refmac's version 5.3.0032
Here go some selected fragments of a 0032 log (no confusion with aniso atomic displacement factors):
>...
>
> ###############################################################
> ###############################################################
> ###############################################################
> ### CCP4 6.0: Refmac_5.3.0032 version 5.3.0032 : 02/16/06##
> ###############################################################
> User: alebus Run date: 8/ 5/2007 Run time: 10:57:27
>...
>Data line--- refi type REST resi MLKF meth CGMAT bref ISOT
> REFI
> Data line--- ncyc 10
> NCYC
> Data line--- scal type BULK reso 2.100 39.621 LSSC ANISO EXPE
> SCAL
> Data line--- solvent YES VDWProb 1.2 IONProb 0.8 RSHRink 0.8
> SOLV
>...
>Cell from mtz : 34.743 124.050 138.303 90.000 90.000 90.000
>Space group from mtz: number - 19; name - P 21 21 21
>...
>-------------------------------------------------------------------------------
> Restraint type N restraints Rms Delta Av(Sigma)
>Bond distances: refined atoms 3149 0.018 0.022
>Bond distances: others 2177 0.002 0.020
>-------------------------------------------------------------------------------
>
>-----------------------------------------------------------------------------
>Overall : scale = 1.143, B = -0.029
>Babinet"s bulk solvent: scale = 0.133, B = 229.342
>Partial structure 1: scale = 0.404, B = 35.602
>Overall anisotropic scale factors
> B11 = 0.00 B22 = 0.00 B33 = 0.00 B12 = 0.00 B13 = 0.00 B23 = 0.00
>-----------------------------------------------------------------------------
>
>-----------------------------------------------------------------------------
>Overall R factor = 0.2092
>Free R factor = 0.2497
>Overall figure of merit = 0.8314
>-----------------------------------------------------------------------------
> Trying gamma equal 0.0000000E+00
> Gamma decreased to 0.0000000E+00
>...
Indeed, now I've updated to Refmac_5.3.0040, and the problem is solved (Rfactors, as expected, are a bit better now; same input, last cycle's figures :
>-------------------------------------------------------------------------------
> Restraint type N restraints Rms Delta Av(Sigma)
>Bond distances: refined atoms 3149 0.017 0.022
>Bond distances: others 2177 0.002 0.020
>-------------------------------------------------------------------------------
>
>-----------------------------------------------------------------------------
>Overall : scale = 1.154, B = -0.003
>Babinet"s bulk solvent: scale = 0.169, B = 300.000
>Partial structure 1: scale = 0.403, B = 43.715
>Overall anisotropic scale factors
> B11 = 0.13 B22 = 1.51 B33 = -1.64 B12 = 0.00 B13 = 0.00 B23 = 0.00
>-----------------------------------------------------------------------------
>
>-----------------------------------------------------------------------------
>Overall R factor = 0.2046
>Free R factor = 0.2424
>Overall figure of merit = 0.8444
>-----------------------------------------------------------------------------
).
I believe version 5.3.0032 was the first announced release when going from 5.2 to 5.3, at the time the very useful keyword 'anom formfactor' was added among other changes (but maybe I'm wrong here...).
Anyway, of course we should be ourselves aware of keeping updated with the latest versions of the software that we use. A lesson for me as well, is to look carefully at the whole output of the programs, not overlooking parameters/figures that we tend to leave as 'unimportant' when they certainly are (even if it's not the Rfac, which tends to become _the_ figure to look at in refinement...).
It might be useful for others though, to check on the Refmac version they're now running and see if these messages are pertinent to them.
cheers,
Ale
On 20/09/2007, Eleanor Dodson <[log in to unmask]> wrote:
> I cant answer - but this version certainly generates anisotropic scales..
>
> We have ccp4.6.0.2 installed)
>
>
> ###############################################################
> ###############################################################
> ###############################################################
> ### CCP4 6.0: Refmac_5.3.0037 version 5.3.0037 : 03/29/06##
> ###############################################################
> User: eleanor Run date: 18/ 9/2007 Run time: 09:34:20
> .....
> -----------------------------------------------------------------------------
> Overall : scale = 0.480, B = -7.297
> Partial structure 1: scale = 0.383, B = 84.682
> Overall anisotropic scale factors
> B11 = -1.37 B22 = -1.37 B33 = 2.06 B12 = -0.69 B13 = 0.00 B23 = 0.00
> -----------------------------------------------------------------------------
>
> You havent got a cubic spacegroup have you?
>
> Eleanor
>
>
> Alejandro Buschiazzo wrote:
> >Hello,
> >I wonder if there was any direct answer to the question that Lari made
> >a couple of days ago.
> >Sorry if I missed it, but I believe there wasn't any response to the
> >ccp4bb...
> >
> >thanks
> >
> >Alejandro
> >
> >
> >Lari Lehtio wrote:
> >>Dear All, I just came back from the beam line and started to
> >>continue my work on a structure that I solved while collecting other
> >>data. I noticed that my R-factors were a lot higher at home lab
> >>although I was using the exact same input files. I then noticed
> >>that the version was different: The one giving lower R-factors
> >>was Refmac_5.2.0019 version 5.2.0019 : 06/09/05 And the one
> >>giving higher: Refmac_5.3.0032 version 5.3.0032 : 02/16/06 I
> >>happened to have the older version also in my home pc, so I tried
> >>again and indeed, 5.2.0019 gives lower R-factors. The input
> >>command is identical, but when I "diffed" the pdb-files, I noticed
> >>that anisotropic scale is missing in the newer files. Did I make a
> >>mistake? How can I turn it back on.
> >>Below is the diff output and the command used for refmac.
> >>Thank you in advance,
> >>~L~
> >>< REMARK 3 PROGRAM : REFMAC 5.2.0019
> >>>REMARK 3 PROGRAM : REFMAC 5.3.0032
> >>< REMARK 3 R VALUE (WORKING + TEST SET) : 0.21101 < REMARK
> >>3 R VALUE (WORKING SET) : 0.20867 < REMARK 3 FREE R
> >>VALUE : 0.25559
> >>>REMARK 3 R VALUE (WORKING + TEST SET) : 0.23077 REMARK 3
> >>>R VALUE (WORKING SET) : 0.22807 REMARK 3 FREE R
> >>>VALUE : 0.28228
> >>< REMARK 3 BIN R VALUE (WORKING SET) : 0.240
> >>>REMARK 3 BIN R VALUE (WORKING SET) : 0.262
> >>< REMARK 3 BIN FREE R VALUE : 0.288
> >>>REMARK 3 BIN FREE R VALUE : 0.343
> >>< REMARK 3 MEAN B VALUE (OVERALL, A**2) : 25.271
> >>>REMARK 3 MEAN B VALUE (OVERALL, A**2) : 29.941
> >>< REMARK 3 B11 (A**2) : 2.35 < REMARK 3 B22 (A**2) :
> >>-1.99 < REMARK 3 B33 (A**2) : 0.32
> >>>REMARK 3 B11 (A**2) : 0.00 REMARK 3 B22 (A**2) :
> >>>0.00 REMARK 3 B33 (A**2) : 0.00
> >>< REMARK 3 B13 (A**2) : 0.90
> >>>REMARK 3 B13 (A**2) : 0.00
> >>
> >>_____________________________________________________________________________________
> >>make check NONE make - hydrogen YES - hout NO - peptide NO -
> >>cispeptide YES - ssbridge YES - symmetry YES - sugar YES -
> >>connectivity NO - link NO refi - type REST - resi MLKF -
> >> meth CGMAT - bref ISOT ncyc 10 scal - type SIMP - reso 2.100
> >>19.976 - LSSC - ANISO - EXPE solvent YES - VDWProb 1.4 -
> >>IONProb 0.8 - RSHRink 0.8 weight - MATRIX 0.1 monitor MEDIUM -
> >>torsion 10.0 - distance 10.0 - angle 10.0 - plane 10.0 -
> >>chiral 10.0 - bfactor 10.0 - bsphere 10.0 - rbond 10.0 -
> >> ncsr 10.0 labin FP=FP SIGFP=SIGFP - FREE=FreeRflag labout FC=FC
> >>FWT=FWT PHIC=PHIC PHWT=PHWT DELFWT=DELFWT PHDELWT=PHDELWT FOM=FOM PNAME
> >>unknown DNAME unknown130907 RSIZE 80 END
> >>_______________________________________ Lari Lehtiö Structural
> >>Genomics Consortium Medical Biochemistry & Biophysics
> >>Dept. Karolinska Institute Stockholm, Sweden
> >>_______________________________________
> >>
> >
> >
--
Alejandro Buschiazzo
Unite de Biochimie Structurale - Institut Pasteur
25, rue du Dr. Roux - 75724 Paris CEDEX - France
Phone: (33) 1 40 61 39 84 Fax: (33) 1 45 68 86 04
|