Before worrying at the rmsbonds check whether you have some significant
outliers - these would be listed in the
REFMAC log file. Sometimes these can really distort the quoted values.
Eleanor
Anastassis Perrakis wrote:
> Hi Rafael,
>
>
>>
>> Things very different on the "new" version:
>> - The rmsANGL shot up (from 2.088 to 3.088);
>> - The rmsBOND also shot up (from 0.25 to 0.0468).
>>
>> As a consequence of having my geometric constrains blown up, the Rwork
>> went considerably down.
>>
>> Does anybody have any thoughts on why using the same geometry restraints
>> led to very different results?
>
> From what you show, the logs are not dramatically different anyway. In
> one case Rfree is 20.37% and in the other 20.4%.
> Which tells you that loosening geometry is not helping ;-)
>
> What you say about the geometry restraints being the same is not correct.
> The relative weight of geometry terms is the same.
> ***But you do not mention anything about the relative weight of Xrays
> to Geometry, which YOU have to set***
> (new REFMAC has an AUTO version which works OK, but not for the very
> final rounds, in my opinion)
>
> Would I be right to suspect you have either set the X-ray vs Geometry
> weight to the wrong value in the old job (0.3 as default?)
> while in the new job AUTO is used?
>
> The rms distance you get in the new job is 0.0468 and the z score is
> 1.958, clearly indicating that the geometry is far too loose.
> In the old job, you had 0.025 and 2.088, which I also think is loose.
> (unless you have VERY high resolution .. and even then ... I would
> expect a better zBOND).
>
> Aim for tighter geometry and look at Z-bond (0.5 - 1.0) are good
> values I think. I personally decided
> I prefer the Ian Tickle analysis btw, and I like looking at the
> weightB relationship to weightXray.
>
> It was early in the year to have the RMS-bonds discussion... (same as
> Jim pointed out for the I/sig(I) discussion).
>
>
> Best, Tassos
>
> PS This script can be useful to decide what good weight are (need
> editing)
> btw, I would love to have that type of Job in ccp4i, and its rather
> easy to code in Tcl.
> Anybody with time to spare on the CCP4 team ? I think it can be useful
> to have a job
> for optimizing weights extensively (phenix has also that mode and I
> like it a lot ... even if its slow).
>
>
>
> #!/bin/csh -f
> #
> echo "xweight bweight rfactor rfree rmszbonds" > stats.log
>
> foreach xweight ( 0.04 0.08 0.12 0.16 0.20 0.25 0.30)
> foreach bweight ( 0.1 0.2 0.6 1.0 1.5 2.0 2.5 3.0 )
> #
> # Define your Input / Output files
> #
> refmac5 XYZIN "CLIC3_final_2.pdb" XYZOUT "del.tmp" HKLIN
> "clic3.mtz" HKLOUT "del.mtz.tmp" TLSIN
> "tls_cli3.tls" TLSOUT "del.tls.tmp" LIBOUT "delete.cif" >
> refmac_${xweight}_${bweight}.log <<EOF
> make check NONE
> make -
> hydrogen ALL -
> 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
> refi tlsc 10
> temp set 20.0
> ncyc 10
> scal -
> type SIMP -
> reso 2.210 38.691 -
> LSSC -
> ANISO -
> EXPE
> solvent YES -
> VDWProb 1.2 -
> IONProb 0.8 -
> RSHRink 0.8
> weight -
> MATRIX ${xweight}
> 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=F_Unspecified SIGFP=SIGF_Unspecified -
> FREE=FreeR_flag
> labout FC=FC FWT=FWT PHIC=PHIC PHWT=PHWT DELFWT=DELFWT
> PHDELWT=PHDELWT FOM=FOM
> temp ${bweight} 2.0 3.0 2.0 3.0
> PNAME fox
> DNAME 040808
> RSIZE 80
> USECWD
> END
> EOF
>
> echo -n ${xweight} ${bweight} >> stats.log
> awk '(NF=11){if ($1==20) print $2, $3, $8}'
> refmac_${xweight}_${bweight}.log >> stats.log
> end
>
> end
>
>
|