> I would like to know the accuracy (error) of the position of the
coordinates
This is provided by the output of the refinement software.
An example (CNS-refined structure, from the pdb header, there is an
input file called "xtal_pdbsubmission.inp"
REMARK 3 ESTIMATED COORDINATE ERROR.
REMARK 3 ESD FROM LUZZATI PLOT (A) : 0.36
REMARK 3 ESD FROM SIGMAA (A) : 0.45
REMARK 3 CROSS-VALIDATED ESTIMATED COORDINATE ERROR.
REMARK 3 ESD FROM C-V LUZZATI PLOT (A) : 0.49
REMARK 3 ESD FROM C-V SIGMAA (A) : 0.60
Another example, from Phenix:
REMARK 3 ERROR ESTIMATES.
REMARK 3 COORDINATE ERROR (MAXIMUM-LIKELIHOOD BASED) : 0.41
A third example, from Refmac:
REMARK 3 ESTIMATED OVERALL COORDINATE ERROR.
REMARK 3 ESU BASED ON R VALUE (A): 0.356
REMARK 3 ESU BASED ON FREE R VALUE (A): 0.279
REMARK 3 ESU BASED ON MAXIMUM LIKELIHOOD (A): 0.282
REMARK 3 ESU FOR B VALUES BASED ON MAXIMUM LIKELIHOOD (A**2): 12.096
Can't provide you with an example for Buster because I requested the
software last year and I do not know what happened to that request.
The coordinate errors are estimated using statistical methods. Relevant
papers to read are those of Randy Read and of Vittorio Luzzatti (and
probably of others), Luzzatti V. (1952), Acta Cryst. 5, 802-810; Read,
R.J. (1986), Acta Cryst. A42, 140-149. There are probably papers by "the
Russian school" (Lunin and others) but I can't locate them right now...
Fred.
|