*** For details on how to be removed from this list visit the *** *** CCP4 home page http://www.ccp4.ac.uk ***
Dear Yingjie Peng, the R-factors look typical to me for a reasonably good MR solution. You don't show the bulk solvent correction parameters of REFMAC, but for the CNS parameters, I would rather fix them to physically more reasonable values, such as electron density ~0.33/0.42 for low salt/high salt buffer and B ~50 A^2, or use a different low resolution cutoff. Your data set probably contains some very low resolution reflections with too low intensities (partial shadow from beam stop or overloaded and poorly extrapolated), leading to a combination of a too high bulk solvent density and too low bulk solvent B-factor (even negative as in your case). Slightly changing the low resolution cutoff to exclude the very lowest reflections might already cure your problem (say from 47 A to 40 A, or so). Model building and refinement will be very difficult at 3.8 A, though. Good luck anyway, Dirk. Yingjie Peng wrote: > *** For details on how to be removed from this list visit the *** > *** CCP4 home page http://www.ccp4.ac.uk *** > > > Dear all, > > > I have collected one 3.8A dataset. The dataset was indexed and scaled as P222, > determined as P212121 using PHASER because the MR using P212121 gave most > significant > result. The cell parameters are a=59.535, b=107.003, c=153.554. > > The PHASER gave me a good MR result (as following). > > ******************************************************* > PHASER result: > > Fast Translation Function Table: Space Group P 21 21 21 > ------------------------------------------------------- > #SET #TRIAL Top (Z) Second (Z) Third (Z) Ensemble > 1 1 2010.62 (35.68) - - - - > > Refinement Table: Space Group P 21 21 21 > ---------------------------------------- > #+ = input number #* = output number > Unsorted (refinement order) Sorted in LL-gain order > Initial Refined Initial Refined > #+ LL-gain LL-gain Unique #* #+ LL-gain LL-gain Unique =#* > =#+ > 1 2139.13 2139.13 YES 1 1 2139.13 2139.13 YES > > ******************************************************* > > But when I tried to do refinement, I felt confused. I have tried both Refmac5 > in > CCP4 and CNS to do refinement. Some results are listed below. > > ******************************************************* > Refmac5 in CCP4i result: > 1. > rigid body refinement using no prior phase information input; > using the pdb file from PHASER as input pdb file; > resolution range: 50.00-3.81; > use matrix scaling; diagonal weighting term 0.001 (or 0.5); > refine overall B-factor (or not) > > give similar results: > R=0.356, Rfree=0.368; > Mean B value (overall)=69.0, with most B fator normal, but > extremely high (>120, even upto 200) for some region. > > > 2. > restrained refinement using no prior phase information input; > using the rigid body refined pdb file from Refmac5 as input pdb file; > resolution range: 47.14-3.81; > use matrix scaling; diagonal weighting term 0.001; > refine overall temperature factors: yielding R=0.37, Rfree=0.39; > refine isotropic temperature factors: yielding R=0.39, Rfree=0.42; > refine anisotropic temperature factors: yielding R=0.38, Rfree=0.41; > refine mixed temperature factors: yielding R=0.39, Rfree=0.42; > > ******************************************************* > CNS result: > using the pdb file from PHASER as input pdb file; > after rigid body refinement, the R=0.40, Rfree=0.40, using initial B-factor > correction and bulk solvent correction, yielding following information, > B-factor correction applied to coordinate array B: -66.713 > solvent: density level= 0.447787 e/A^3, B-factor= 146.09 A^2 > > after minimizing, using initial B-factor correction and bulk solvent > correction, > the R=0.35, Rfree=0.48, with following information, > B-factor correction applied to coordinate array B: 4.669 > bulk solvent: density level= 0.454096 e/A^3, B-factor= 182.182 A^2 > > I have tried different protocols. Minimizing without initial B-factor > correction > but with bulk solvent correction did NOT make big difference, but minimizing > without initial B-factor correction and also without bulk solvent correction > yielding a higher R (0.37) but a lower Rfree (0.46). > ******************************************************* > > > I am wondering the initial B-factor correction and bulk solvent correction in > CNS play what kind of role in the refinement. Can these corrections be applied > to low resolution dataset, like my 3.8A dataset? If yes, how to use them > correctly? > > Thank you in advance. > > Best wishes, > > Yingjie Peng > > > Yingjie PENG, Ph.D. student > Structural Biology Group > Shanghai Institute of Biochemistry and Cell Biology (SIBCB) > Shanghai Institute of Biological Sciences (SIBS) > Chinese Academy of Sciences (CAS) > 320 Yue Yang Road, Shanghai 200031 > P. R. China > 86-21-54921217 > Email: [EMAIL PROTECTED] > > -- **************************************** Dirk Kostrewa Paul Scherrer Institut Biomolecular Research, OFLC/110 CH-5232 Villigen PSI, Switzerland Phone: +41-56-310-4722 Fax: +41-56-310-5288 E-mail: [EMAIL PROTECTED] http://sb.web.psi.ch ****************************************
