I do think at times the Rfree is a bit of a sacred cow . At one level if the maps look OK not much can be wrong.. Is the Ramachandran plot OK?
You dont say what the R factor is but does this mean there is a large difference between the Rfactor and Rfree? If you are using REFMAC look at the plots of R factor and Rfree v resolution? Is there a big divergence at some points? Could you have included overloads? Are there icerings? Then look at the plot of <Fos> v <Fcalc> v resolution/ Is the scaling procedure meaning these diverge in some regions? Is there some strange effects at low resolution? And so on and so on.. Eleanor On Tue, 9 Oct 2018 at 19:31, Wim Burmeister <wim.burmeis...@ibs.fr> wrote: > Hello, > at that resolution, the refinement of anisotropic atomic B-factors is > absolutely required, as is the modelisation of alternate conformations for > surface residues. Optimize also the weight of different restraints (for > exemple on B-factors) in order to get the lowest Rfree. > Best > Wim > > ------------------------------ > *De: *"Robbie Joosten" <robbie_joos...@hotmail.com> > *À: *CCP4BB@JISCMAIL.AC.UK > *Envoyé: *Mardi 9 Octobre 2018 20:02:17 > *Objet: *Re: [ccp4bb] Issue with high Rfree (0.25) for a high-resolution > dataset (1.05 Ang) > > Hi Hugo, > > Perhaps you should play with your refinement strategy. Use a decent number > of cycles and a sensible restraint weight (something that gives you rmsZ < > 1.0 and good R-factors). Anisotropic B-factors are probably needed and make > your model as complete as your maps allow. > > You could try pdb-redo to see if this can help you on your way. > > Cheers, > Robbie > > On 9 Oct 2018 19:12, Guto Rhys <guto.r...@bristol.ac.uk> wrote: > > Hi all, > > I have a 1.05 Angstrom dataset that I was able to phase but the refined > model only has an Rfree of approximately 0.25. The dataset includes 1800 > images and, as the crystal did not suffer significantly from radiation > damage, comprises all 360 deg. Auto-processing pipelines at diamond light > source all suggest I222. I have also indexed the data in iMOSFLM, which has > the highest-symmetry Laue group that is least penalised of I422. Subsequent > scaling and merging in AIMLESS strongly indicates that I222 is the likely > space group (see below). I have ran the refined model through ZANUDA, which > has similar R values to lower symmetry space groups (see below). The output > from Phenix Xtriage does not find any specific crystal pathologies and if > twinning is present it is very low (2 to 4%, see below). The difference map > suggests that the model accounts for nearly all the density. Any ideas or > direction would be greatly appreciated. > > Best, > Guto > > > AIMLESS Summary > Overall InnerShell OuterShell > Low resolution limit 27.75 27.75 1.07 > High resolution limit 1.05 5.75 1.05 > > Rmerge (within I+/I-) 0.050 0.078 0.466 > Rmerge (all I+ and I-) 0.051 0.080 0.536 > Rmeas (within I+/I-) 0.055 0.086 0.591 > Rmeas (all I+ & I-) 0.054 0.085 0.613 > Rpim (within I+/I-) 0.023 0.034 0.359 > Rpim (all I+ & I-) 0.017 0.028 0.288 > Rmerge in top intensity bin 0.049 - - > Total number of observations 107950 779 1972 > Total number unique 11315 88 486 > Mean((I)/sd(I)) 19.7 46.2 1.8 > Mn(I) half-set correlation CC(1/2) 0.998 0.994 0.796 > Completeness 99.1 99.2 90.4 > Multiplicity 9.5 8.9 4.1 > > Anomalous completeness 98.1 100.0 79.1 > Anomalous multiplicity 5.0 6.4 2.2 > DelAnom correlation between half-sets -0.067 0.286 0.097 > Mid-Slope of Anom Normal Probability 0.789 - - > > Estimate of maximum resolution for significant anomalous signal = 1.14A, > from CCanom > 0.15 > > Estimates of resolution limits: overall > from half-dataset correlation CC(1/2) > 0.30: limit = 1.05A == > maximum resolution > from Mn(I/sd) > 1.50: limit = 1.05A == > maximum resolution > from Mn(I/sd) > 2.00: limit = 1.07A > > Estimates of resolution limits in reciprocal lattice directions: > Along h axis > from half-dataset correlation CC(1/2) > 0.30: limit = 1.06A > from Mn(I/sd) > 1.50: limit = 1.09A > Along k axis > from half-dataset correlation CC(1/2) > 0.30: limit = 1.11A > from Mn(I/sd) > 1.50: limit = 1.13A > Along l axis > from half-dataset correlation CC(1/2) > 0.30: limit = 1.05A == > maximum resolution > from Mn(I/sd) > 1.50: limit = 1.05A > > Anisotropic deltaB (i.e. range of principal components), A^2: 6.40 > > Average unit cell: 29.12 29.26 55.50 90.00 90.00 90.00 > Space group: I 2 2 2 > Average mosaicity: 0.36 > > AIMLESS Laue Group prediction > > Laue Group Lklhd NetZc Zc+ Zc- CC CC- Rmeas R- > Delta ReindexOperator > > = 1 I m m m *** 0.987 6.25 9.19 2.94 0.92 0.29 0.07 0.49 > 0.0 [h,k,l] > 2 I 1 2/m 1 0.004 4.36 9.32 4.96 0.93 0.50 0.07 0.30 > 0.0 [-h,-k,l] > 3 I 1 2/m 1 0.004 4.14 9.24 5.10 0.92 0.51 0.07 0.30 > 0.0 [k,-h,l] > 4 I 1 2/m 1 0.004 4.05 9.23 5.18 0.92 0.52 0.06 0.31 > 0.0 [h,-l,k] > 5 I 4/m m m 0.000 6.35 6.35 0.00 0.64 0.00 0.22 0.00 > 0.3 [h,k,l] > 6 I 4/m 0.000 0.90 6.83 5.93 0.68 0.59 0.17 0.26 > 0.3 [h,k,l] > 7 P -1 0.000 3.69 9.36 5.67 0.94 0.57 0.06 0.26 > 0.0 [-h,k,1/2h-1/2k-1/2l] > 8 I 1 2/m 1 0.000 0.08 6.40 6.33 0.64 0.63 0.23 0.22 > 0.3 [-1/2h+1/2k-1/2l,-h-k,-1/2h+1/2k+1/2l] > 9 F m m m 0.000 -0.43 6.17 6.60 0.62 0.66 0.24 0.20 > 0.3 [h+k,-h+k,l] > 10 I 1 2/m 1 0.000 0.75 6.89 6.14 0.69 0.61 0.22 0.22 > 0.3 [-1/2h-1/2k-1/2l,h-k,-1/2h-1/2k+1/2l] > > > Xtriage Summary > > Twinning and intensity statistics summary (acentric data): > > Statistics independent of twin laws > <I^2>/<I>^2 : 2.126 (untwinned: 2.0, perfect twin: 1.5) > <F>^2/<F^2> : 0.774 (untwinned: 0.785, perfect twin: 0.885) > <|E^2-1|> : 0.761 (untwinned: 0.736, perfect twin: 0.541) > <|L|>, <L^2>: 0.490, 0.323 > Multivariate Z score L-test: 1.303 > > The multivariate Z score is a quality measure of the given > spread in intensities. Good to reasonable data are expected > to have a Z score lower than 3.5. > Large values can indicate twinning, but small values do not > necessarily exclude it. > > > > Statistics depending on twin laws > ----------------------------------------------------------------------------------- > > | Operator | type | R_abs obs. | R_abs calc. | Britton alpha | H alpha | > ML alpha | > ----------------------------------------------------------------------------------- > > | k,h,-l | PM | 0.469 | 0.478 | 0.016 | 0.041 | > 0.022 | > ----------------------------------------------------------------------------------- > > > Patterson analyses > - Largest peak height : 13.667 > (corresponding p value : 0.04780) > > > The largest off-origin peak in the Patterson function is 13.67% of the > height of the origin peak. No significant pseudotranslation is detected. > > The results of the L-test indicate that the intensity statistics > behave as expected. No twinning is suspected. > > Zanuda output > > Step 2. > Refinements in subgroups. > There are 5 subgroups to test. > > current time: Jan 08 16:20 GMT > expected end of job: Jan 08 16:28 GMT > > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ > | >> 5 | I 2 2 2 | 0.0005 | -- | -- | -- | > --------------------------------------------------------------------- > | 1 | P 1 | 0.0731 | 0.3242 | 0.2842 | 0.2768 | > | 2 | C 1 2 1 | 0.0755 | 0.3796 | 0.2899 | 0.2799 | > | 3 | C 1 2 1 | 0.0727 | 0.3385 | 0.2909 | 0.2806 | > | 4 | C 1 2 1 | 0.0729 | 0.3398 | 0.2900 | 0.2813 | > | 5 | I 2 2 2 | 0.0707 | 0.3582 | 0.2976 | 0.2802 | > --------------------------------------------------------------------- > | << 1 | P 1 | 0.0731 | 0.3242 | 0.2842 | 0.2768 | > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ > > Step 3. > Refinement of the best model. > Candidate symmetry elements are added one by one. > > current time: Jan 08 16:25 GMT > expected end of job: Jan 08 16:28 GMT > > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ > | >> 1 | P 1 | 0.0731 | 0.3242 | 0.2842 | 0.2768 | > --------------------------------------------------------------------- > | 1 | P 1 | 0.0730 | 0.3243 | 0.2842 | 0.2768 | > | 4 | C 1 2 1 | 0.0751 | -- | 0.2896 | 0.2826 | > | 5 | I 2 2 2 | 0.0737 | -- | 0.2970 | 0.2801 | > --------------------------------------------------------------------- > | << 5 | I 2 2 2 | 0.0737 | -- | 0.2970 | 0.2801 | > --------------------------------------------------------------------- > > R-factor in the original subgroup is (almost) the best. > The original spacegroup assignment seems to be correct. > > ######################################################################## > > To unsubscribe from the CCP4BB list, click the following link: > https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB&A=1 > > > > ------------------------------ > > To unsubscribe from the CCP4BB list, click the following link: > https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB&A=1 > > > ------------------------------ > > To unsubscribe from the CCP4BB list, click the following link: > https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB&A=1 > ######################################################################## To unsubscribe from the CCP4BB list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB&A=1
