Dear Seth,

    I'm not Paul, but I have been able to create cif's that are good
enough to allow rotation about bonds.  A cif is not dependent on
spacing so I don't think that is the source of your problem.  I
doubt the quotes are involved either.

    The "chi" angles are encoded in the _chem_comp_tor loop.  The
loop I have in one of my cifs is

loop_
_chem_comp_tor.comp_id
_chem_comp_tor.id
_chem_comp_tor.atom_id_1
_chem_comp_tor.atom_id_2
_chem_comp_tor.atom_id_3
_chem_comp_tor.atom_id_4
_chem_comp_tor.value_angle
_chem_comp_tor.value_angle_esd
_chem_comp_tor.period
  BCL     tor_0001    C2B     C3B     CAB     OBB         0.000    5.000    2
  BCL     tor_0002    C2C     C3C     CAC     CBC        60.000   15.000    3
  BCL     tor_0003    CHA     CBD     CGD     O2D        60.000   15.000    2
  BCL     tor_0004    CBD     CGD     O2D     CED       180.000    5.000    1
  BCL     tor_0005    C1A     C2A     CAA     CBA        60.000   15.000    3
  BCL     tor_0006    C2A     CAA     CBA     CGA        60.000   15.000    3
  BCL     tor_0007    CAA     CBA     CGA     O2A        30.000   15.000    6
  BCL     tor_0008    CBA     CGA     O2A     C1          0.000    5.000    2
  BCL     tor_0009    CGA     O2A     C1      C2         30.000   15.000    6
  BCL     tor_0010    O2A     C1      C2      C3         30.000   15.000    6
  BCL     tor_0011    C1      C2      C3      C5        180.000    5.000    1
  BCL     tor_0012    C2      C3      C5      C6         30.000   15.000    6
  BCL     tor_0013    C3      C5      C6      C7         60.000   15.000    3
  BCL     tor_0014    C5      C6      C7      C8         60.000   15.000    3
  BCL     tor_0015    C6      C7      C8      C10        60.000   15.000    3
  BCL     tor_0016    C7      C8      C10     C11        60.000   15.000    3
  BCL     tor_0017    C8      C10     C11     C12        60.000   15.000    3
  BCL     tor_0018    C10     C11     C12     C13        60.000   15.000    3
  BCL     tor_0019    C11     C12     C13     C15        60.000   15.000    3
  BCL     tor_0020    C12     C13     C15     C16        60.000   15.000    3
  BCL     tor_0021    C13     C15     C16     C17        60.000   15.000    3
  BCL     tor_0022    C15     C16     C17     C18        60.000   15.000    3
  BCL     tor_0023    C16     C17     C18     C20        60.000   15.000    3
  BCL     tor_0024    CGD     O2D     CED     HED1        0.000   20.000    1
  BCL     tor_0025    C3D     C2D     CMD     HMD1        0.000   20.000    1
  BCL     tor_0026    C3C     CAC     CBC     HBC1        0.000   20.000    1
  BCL     tor_0027    C3C     C2C     CMC     HMC1        0.000   20.000    1
  BCL     tor_0028    C3B     CAB     CBB     HBB1        0.000   20.000    1
  BCL     tor_0029    C1B     C2B     CMB     HMB1        0.000   20.000    1
  BCL     tor_0030    C4A     C3A     CMA     HMA1        0.000   20.000    1
  BCL     tor_0031    C2      C3      C4      H41         0.000   20.000    1
  BCL     tor_0032    C7      C8      C9      H91         0.000   20.000    1
  BCL     tor_0033    C12     C13     C14     H141        0.000   20.000    1
  BCL     tor_0034    C17     C18     C19     H191        0.000   20.000    1
  BCL     tor_0035    C17     C18     C20     H201        0.000   20.000    1

    It is possible that the connectivity tree in _chem_comp_tree is also
used in this function.  Writing code to generate this tree was the most
annoying aspect of creating a cif. What does your tree look like?

Dale Tronrud

Seth Harris wrote:
> Hi,
> 
> Can someone (e.g. Paul) tell me what part of the .cif restraints library 
> COOT uses (and what format is expected) to recognize rotatable bonds in 
> small molecules (i.e. Chi angles)?
> 
> I generally use libcheck and corina to generate .cif files of novel 
> compounds in our structures semi-automatically, and while the latter 
> produces more robust information, Coot feels that the ligand has no Chi 
> angles once the library is read (though it does report X number of 
> restraints successfully and the small molecule behaves appropriately 
> during real space refinement). Monomer library sketcher also does not 
> recognize which groups are planar groups but does show the structure. So 
> at the moment I read in libcheck's library so I can spin various bonds, 
> then read in corina's library again so I can snuggle the molecule into 
> density without collapsing various ring systems inappropriately. Ugh.
> 
> Looking at the file contents they have similar loop structures and 
> content with 3 notable differences. My corina examples lack H atom 
> descriptors, have single quotes around many of the data fields, and the 
> field columns are spaced differently. A few quick tests suggest that the 
> last issue may be the crux of the matter. Are these files meant to be 
> hard-wired to exact column spacings, like the lovely PDB format? I 
> thought cif was supposed to get away from such punch-card ilke behavior, 
> and I'm a bit confused since as mentioned the 'wrongly' spaced .cif 
> restraints library is perfectly functional in real-space refinement 
> (coot) or in refmac, just not so for edit chi angles (coot) or 
> displaying planar groups (sketcher). Is there any way Coot (and ccp4, 
> but perhaps all this should be posed on that list?) could be convinced 
> to recognize these aspects? My last request on a different topic was 
> scheduled for v0.9 disappointingly (or perhaps Paul will be skipping 
> 5,6,7, and 8?? Obviously I'm too focused on the small stuff!) so if this 
> one too I believe I'll have to get around to writing a paper, citing 
> coot, and trying to time it to win a citation-based feature request 
> award. Ha! Or I'd write a script to bring the corina file in line...I 
> just need to know what the target actually is. I did make it most of the 
> way through the ccp4 mmcif format web page, but I found my question 
> wasn't going away.
> 
> Thanks as always for the great program, features, and rapid development.
> 
> Cheers,
> Seth