On Thu, Dec 15, 2011 at 9:05 AM, Ed Pozharski
As for R-free flags, did you generate them in phenix? If so, the refmac run with default parameters that pdb does during validation would run against about 5% of reflections (in phenix test_value=1, in refmac it's zero).
I thought the PDB finally fixed this? Their R-factor validation is nearly useless anyway, but this seems like such an obvious and easy mistake to avoid.
I guess general way to avoid such calamity is to generate test set with freeflag initially. At this point, you can definitely use something like sftools to reset the flags so that your model passes pdb validation with flying colors.
Does freerflag take into account the possibility of twinning? FYI, you can also use the reflection file editor in the GUI to convert the flags to something CCP4 can understand, with the caveat that any flags other than '0' in the output file are not guaranteed to be evenly distributed or reproducible. -Nat