J. Pontius et al., DEVIATIONS FROM STANDARD ATOMIC VOLUMES AS A QUALITY MEASURE FOR PROTEIN CRYSTAL-STRUCTURES, Journal of Molecular Biology, 264(1), 1996, pp. 121-136
Standard ranges of atomic and residue volumes are computed in 64 highl
y resolved and well-refined protein crystal structures using the class
ical Voronoi procedure. Deviations of the atomic volumes from the stan
dard values, evaluated as the volume Z-scores, are used to assess the
quality of protein crystal structures. To score a structure globally,
we compute the volume Z-score root mean square deviation (2-score rms)
, which measures the average magnitude of the volume irregularities in
the structure. We find that the Z-score rms decreases as the resoluti
on and R-factor improve, consistent with the fact that these improveme
nts generally reflect more accurate models. From the Z-score rms distr
ibution in structures with a given resolution or X-factor, we determin
e the normal limits in Z-score rms values for structures solved at tha
t resolution or X-factor. Structures whose Z-score rms exceeds these l
imits are considered as outliers. Such structures also exhibit unusual
stereochemistry, as revealed by other analyses. Absolute Z-scores of
individual atoms are used to identify problems in specific regions wit
hin a protein model. These Z-scores correlate fairly well with the ato
mic B-factors, and atoms having absolute Z-scores >3, occur at or near
regions in the model where programs such as PROCHECK identify unusual
stereochemistry. Atomic volumes, themselves not directly restrained i
n crystallographic refinement, can thus provide an independent, rather
sensitive, measure of the quality of a protein structure. The volume-
based structure validation procedures are implemented in the program P
ROVE (PROtein Volume Evaluation), which is accessible through the Worl
d Wide Web. (C) 1996 Academic Press Limited