The question: "I have just finished refining a structure at 2.5 A resolution. On the whole the geometry is very good (RMS bonds = 0.009A, angle distance = 0.037A, etc). The model has a total of 5 pepflip outliers (> 2.5 A) out of 269 residues. Two of these precede cis-prolines (3.62 and 2.82 A) and one is a cis-proline (2.52 A). The structure was refined using PROLSQ and I identified the cis-Prolines in the PROTIN stage. Incidentally, the final 2Fo-Fc map is good for all the outliers and the difference map is essentially flat. Has anyone else experienced this, or is it peculiar to my structure (or my refinement protocol !). Perhaps the database used by the pep_flip algorithm doesn't contain many (or any ?) cis-prolines. Any ideas ?"
Morten replied: "Cis-prolines gives high pep-flip values; this is the normal and desired behaviour."