![](https://secure.gravatar.com/avatar/cad35f4dd77b3ab8e60ede4e3d54c6f4.jpg?s=120&d=mm&r=g)
On Fri, Nov 11, 2011 at 11:05 AM, Green, Todd
If the current refinement in C2 (with twin) were correct, are the numbers r_work = 0.1582 r_free = 0.1866 too low to be reasonable?
They're not unheard of - the attached image shows the distribution of R-free values in the PDB for structures between 2.25 and 2.35 Angstrom resolution, and an R-free of 0.1866 does fall within this distribution. But it's low enough that you should probably be a little suspicious, and check to make sure you're not doing anything that might bias the statistic. (Note that the presence of NCS can also make it very difficult to obtain an unbiased R-free - you can try assigning the flags in thin resolution shells, which Phenix can also do, but this isn't a guaranteed solution.) -Nat