[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [ccp4bb]: infinite loops in refmac?
*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
Well - I downloaded this model and I see why it failed, but not how to
fix it. After the unrestrained refinement the geometry is so bad that
the program gets confuse - there are severe symmetry clashes; bond
errors of several As and so on.
I think you need to tidy the geometry although that seems pretty
challenging! - maybe not to do any unrestrained refinement at this
resolution. The ArpWarp scripts should work beautifully here I would
have thought?
And we will try to find out where that pesky nan comes from!
Best wishes Eleanor
Sample of geometry listing - by the way , when things get ugly always
turn on MONITOR MANY..
WARNING : big distance for conn:TRANS dist = 2.578
ch:XX res: 25 ILE --> 26 ALA ideal_dist=
1.329
WARNING : CIS peptide bond is found, angle = 24.01
ch:XX res: 29 LYS --> 30 TYR
WARNING : big distance for conn:TRANS dist = 2.476
ch:XX res: 30 TYR --> 31 GLU ideal_dist=
1.329
WARNING : CIS peptide bond is found, angle = 8.63
ch:XX res: 31 GLU --> 32 THR
WARNING : big distance for conn:PTRANS dist = 3.019
ch:XX res: 32 THR --> 33 PRO ideal_dist=
1.329
WARNING : big distance for conn:TRANS dist = 2.200
Mark DePristo wrote:
> Hello everyone,
>
> I've been experiencing some odd behavior with refmac5. On some input
> files, refmac5 will enter into something like an infinite loop, never
> terminating. Things start to go bad from the start (the nan values for
> bond length and angle RMS deltas):
>
> CGMAT cycle number = 1
>
>
> -------------------------------------------------------------------------------
>
> Restraint type N restraints Rms Delta Av(Sigma)
> Bond distances: refined atoms 827 1.891 0.022
> Bond angles : refined atoms 1124 nan 1.942
> -------------------------------------------------------------------------------
>
>
>
> -----------------------------------------------------------------------------
>
> Overall : scale = 1.014, B = -1.043
> Babinet"s bulk solvent: scale = 0.764, B = 50.000
> Partial structure 1: scale = 0.143, B = 30.515
> Overall anisotropic scale factors
> B11 = 0.01 B22 = -0.09 B33 = 0.07 B12 = 0.00 B13 = -0.01 B23 = 0.00
> -----------------------------------------------------------------------------
>
>
>
> -----------------------------------------------------------------------------
>
> Overall R factor = 0.4249
> Free R factor = 0.5012
> Overall figure of merit = 0.4024
> -----------------------------------------------------------------------------
>
>
>
>
>
>
>
>
>
> CGMAT cycle number = 2
>
>
> -------------------------------------------------------------------------------
>
> Restraint type N restraints Rms Delta Av(Sigma)
> Bond distances: refined atoms 827 nan 0.022
> Bond angles : refined atoms 1124 nan 1.942
>
> The program never gets past this last line. The input PDB file is the
> result of 20 steps of unrestrained refinement against a 1.31 Å dataset
> with refmac5. The behavior is simple to reproduce, just run:
>
> .../refine refine-2-1.pdb refine-2-1.mtz RESTrained 20 output.pdb
> output.mtz weak isotropic 0.75
>
> meaning approximately to refine refine-2-1.pdb against refine-2-1.mtz
> with 20 steps of restrained refinement with weight matrix 0.75, weak
> restraints and isotropic b-factors.
>
> The refine script and input files can be found at:
>
> http://raven.bioc.cam.ac.uk/~mdepristo/refmac/refine-2-1.pdb
> http://raven.bioc.cam.ac.uk/~mdepristo/refmac/refine-2-1.mtz
> http://raven.bioc.cam.ac.uk/~mdepristo/refmac/refine
>
> I've had this crop up a couple of times already and it's starting to
> become a serious nuisance. Has anyone else experienced this? Is there a
> workaround?
>
> Thanks,
>
> Mark
>
> Mark DePristo
> Ph.D. Candidate
> Dept. of Biochemistry
> Cambridge University
> mdepristo@cryst.bioc.cam.ac.uk
> http://www-cryst.bioc.cam.ac.uk/~mdepristo/