 |
 |
I dont think REFMAC idealises the Ramachandran plot in any way - the theory is this geometric check should be kept as an independent validation tool. You can use COOT to idealise the Ramachandran torsion angles if you wish, but it would be sensible to check against the electron density. Eleanor On 5 February 2016 at 11:53, Quyen Hoang <[log in to unmask]> wrote: > As Bernhard said, look at the density around each outliers. But keep in > mind that torsions outside of the 'favorable' regions on the Ramachandran > map do exist, especially of residues in the active-site of enzymes. > > Cheers > Quyen > > Sent from my BlackBerry 10 smartphone. > *From: *Bernhard Rupp (Hofkristallrat a.D.) > *Sent: *Friday, February 5, 2016 6:40 AM > *To: *[log in to unmask] > *Reply To: *[log in to unmask] > *Subject: *Re: [ccp4bb] on structure idealization by refmac5 > > As always, look at the map. Ramachandran errors mean that the geometric > crossvalidation > > (unrestrained backbone torsions phi psi) tells you that something is not > quite right. > > > > These errors therefore most frequently appear when you have nothing (in > terms of electron density or > > scattering contribution) that holds your unrestrained backbone geometry > > (in contrast to the fairly restrained stereochemistry ones like bonds, > angles etc) in check. The refinement > > program does not want the atoms there, it can’t send them into space (like > it can with waters), has run out of > > increasing B-factor, and then the atoms start to shift around meaningless, > at the expense of plausible > > backbone geometry, and implausible high energy conformations appear. > > > > That is the value of Ramachandran analysis. It tells you to look at the > map and figure out what is wrong. > > Often one just has to admit that evidence is insufficient to support a > poor model at that particular > > proposed location. > > > > Best, BR > > > > *From:* CCP4 bulletin board [mailto:[log in to unmask]] *On Behalf Of *Smith > Liu > *Sent:* Freitag, 5. Februar 2016 12:19 > *To:* [log in to unmask] > *Subject:* [ccp4bb] on structure idealization by refmac5 > > > > Dear All, > > If after structure idealization by refmac 5, the ratio of Ramachandran > favored of a model increase from 88% to 91%, does this mean the model may > have significant errors at least in some parts which hinders further > idealization by refmac5, as for a good model usually has Ramachandran > favoured ratio higher than 95%? Is anyway which can increase the ratio of > Ramachandran favoured further, to the level of most idealized model? > > Smith > > > > > >