Typically, if clashes don't get resolved automatically it means bits of model where clashes occur are trapped in local minima that cannot be escaped by gradient-driven minimization. Try SA refinement or fix manually, then re-run your usual refinement. Pavel On 9/27/15 18:21, Smith Lee wrote:
Dear All,
With the EM cryo data, I have compared the function of phenix.real_space_refine of Phenix10 and Phenix1.9-1692. For the Ramachandran favored, the new version is about 90%, the old version is about 92%. For the clashscore by Molprobity, the new version is about 30, the old version is about 20. Molprobity indicated the clashing atoms were systematically existing in the whole molecule, rather than on specific parts of the protein which may be indication of bad modelling.
Clearly the clashscore of 30 is unacceptable. Will you please give some advise on how to improve the function of new version of Phenix, considering adding H was not suitable as for the resolution was poorer than 4.0 A?
Smith
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb Unsubscribe: [email protected]