Re: [PyMOL] Bug in interpretation of SCALEn records

[Thomas H. <tho...@sc...>]

Hi Takanori,

PyMOL uses SCALEn to transform into fractional coordinates, and then the
unit cell parameters (CRYST1) to transform back into cartesian
coordinates. The PDB spec says: "The unit cell parameters are used to
calculate SCALE". For 1VW3, CRYST1 defines a unit cube, so I assume
SCALEn should be the identity matrix, but it's not.

You can skip this back-and-forth transformation with a setting:

PyMOL> set pdb_insure_orthogonal, off

I assume PyMOL is doing the right thing here. If we want to know more
about the intended interplay of CRYST1, ORIGXn and SCALEn, we might want
to crosspost the topic to the CCP4BB list.

Cheers,
  Thomas

Takanori Nakane wrote, On 05/24/14 03:48:
> Dear all,
> 
> I noticed a strange behaviour in the handling of SCALEn
> records in PDB files.
> 
> When I load 1VW3, all atomic coordinates are scaled down
> to one thousandth and therefore all atoms get connected.
> The fact that the coordinates are incorrectly transformed
> can be confirmed by saving the loaded model again.
> 
> This is probably due to the presence of SCALEn records:
> 
>> SCALE1      0.001000  0.000000  0.000000        0.00000                         
>> SCALE2      0.000000  0.001000  0.000000        0.00000                         
>> SCALE3      0.000000  0.000000  0.001000        0.00000 
> 
> When they are deleted, the model can be loaded correctly.
> 
> According to the PDB format specification
> http://deposit.rcsb.org/adit/docs/pdb_atom_format.html#SCALEn ,
> SCALEn records should not affect real space (orthogonal) coordinates.
> 
> Best regards,
> 
> Takanori Nakane

-- 
Thomas Holder
PyMOL
Schrödinger, Inc.