Dear Eleanor, Ros,
On 29/04/12 11:32, Eleanor Dodson wrote:
I think you will find the dictionaries for coot and refmac are
different..
REFMAC default dictionary will $CLIBD/monomers/n/NAD.cif
Who knows where coot finds its dictionaries ...
Coot uses the environment variable
COOT_REFMAC_LIB_DIR
if that is not set, it uses
COOT_MONOMER_LIB_DIR
if that is not set, it uses
CLIBD_MON
if that it not set, it uses
CCP4_LIB
if that is not set, it uses
COOT_PREFIX/share/coot/lib
which means that Coot typically picks up the standard CCP4 library, but
allows you to over-ride that for special cases, or if you are using Phenix.
So it is good advice to check for a start that Coot and Refmac are using
the same restraints. When Coot reads in the mmCIF dictionary, it writes
that information to the console.
1) check the REFMAC restraints in that dictionary are sensible -
spec, are the planes and chiral centres as you would expect
It is my experience that Refmac restraints used in Coot tend to flatten
furanoses and pyrrolidines.
2) read the REFMAC dictionary into coot and see what happens.. do you
still have the problem.
3) check the REFMAC log file for what it thinks are outliers after the
COOT refinement - I reset the geometric monitoring parameters - see on
the GUI - to list things with errors of 5 Sigma say ( I think the
default is 10 Sigma)
Yes, agreed.
4) If there are difference complain..
And if you do so, be sure to tell us which versions of refmac and coot
you are using - we find that sort of thing important when debugging.
Also, in this case, it might be useful to know about the restraints
weights, presumably the default seeing as you don't mention it?
BTW, I rarely if ever use Regularize Zone if there electron density to fit.
If you hover over the red bar in the geometry analysis in Coot, it tells
you the most deviant geometrical feature. Is this consistent with the
analysis from Refmac?
Regards,
Paul.
