***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***


This looks perfectly reasonable in terms of initial numbers, so my questions would be: what did you do to the model before you started refining ? Did you delete all the obviously different bits in with the existing sequence ? Strip to ALA and keep only the identical residues ? Refine B-factors after energy minimization ?

At 24% identity at 2.4 Angstrom, I'm not surprised that your R-free goes up during refinement unless you've made particularly shrewd changes to your model but you'd still hope that the (R-free - R) difference would be less than 8% and the geometry would be tight (0.010 or tigher RMS bond length deviations). At 44% or 49% R-free your maps aren't going to look that good. I'd suggest running one cycle of energy minimization and fairly heavily restrained B-factor refinement, calculate maps, then chop out any loops that aren't in reasonable density. I'd look at the hydrophobic core(s), the areas with lowest regional average B-factor, and start making changes then the work outwards. Change at most 10% of the structure, refine again, recalculate maps, lather rinse repeat. It can be tedious but the risk at this stage is that if you make too many changes in a marginal map you can bury yourself in a wrong interpretation.

At this stage I'm not sure whether to advocate chopping non-identical residues back to ALA or not. Not to do so risks burying yourself in a fairly deep pit of phase bias. To do so might actually make the phases worse - I've had two recent fairly tough-ish cases where that seemed to be the case and they were around this sequence identity range.

Good luck,
Phil Jeffrey

Crystallography Facility Manager
Dept Molecular Biology
Princeton University

Reply via email to