In the case of anisotropic protein model refinement undertaken at a diffraction resolution worse than ~1.6 Angstrom the DPI formula (2a) denominator value of p (Where p = Number of observations - Number of refined parameters ) can approach zero and the DPI estimate thereby becomes unstable. Please consider carefully why you are undertaking anisotropic refinement at such a diffraction resolution.

The case of PDB files including hydrogens at calculated positions approximately doubles the number of atoms and thereby affects the Diffraction Precision Index (DPI) value. However since such stereochemically derived positions are not experimentally determined from the diffraction data the DPI formula for the number of atoms needs to be the non-hydrogen atoms number only. This has been corrected for this specific case on March 11, 2016.