Others have suggested sensible matrix values but do remember that initial "refinement" after MR is not meant to be producing a final model with good geometry - you are using it to sort out the correct parts of your solution from the junk, and the overall geometry parametrs are not very important. At the end of the first pass with a FreeR > 40% you should have some residues which are clearly wrong; bad geometry and contacts, high B factors etc. I usually remove those from the model at that stage then rebuild them using the phase information from the good bits Eleanor
[EMAIL PROTECTED] wrote: >*** For details on how to be removed from this list visit the *** >*** CCP4 home page http://www.ccp4.ac.uk *** > > >Thanks for all kind replies, I changed the matrix value to 0.05, and it works. > >Best regards, > >Yuhe Liang, PhD >Structural Biology Lab >College of Life Sciences >Peking University >Beijing, 100871 >P.R.China >Tel: 86-10-62755714 >Fax: 86-10-62765669 > >=======In your last mail at 2005-12-02 11:42:00 ======= > > > >>*** For details on how to be removed from this list visit the *** >>*** CCP4 home page http://www.ccp4.ac.uk *** >> >> >> >>On Dec 2, 2005, at 10:21, Tim Gruene wrote: >> >> >> >>>*** For details on how to be removed from this list visit the *** >>>*** CCP4 home page http://www.ccp4.ac.uk *** >>> >>> >>>Hello, >>> >>>you have to lower the weight between crystrallographic and geometric >>>data. >>>That is the matrix keyword. I think the default is 0.3, which in my >>>experience is much to high at the beginning of refinement. >>> >>> >>Its completely resolution dependent. >> >>MATRIX 0.3 is correct for about 2.0 A. >>for 2.5 A you sometimes need 0.05. >>for 1.5 A, 0.5-0.7 tends to be OK. >> >>I would suggest to aim for 1-2 distances rms between 0.015-0.020 and >>angles about 1.5 degrees. >> >> >> >> >>>You can try >>>setting it to AUTO instead of a number and let refmac figure it out for >>>you. If the protein is still distorted, lower the value until the >>>geometric statistics look reasonable. >>> >>> >>> >>AUTO is great as a rough guess - in my experience towards the end of >>refinement you better tune yourself. >> >> Tassos >> >> >> >>>Tim >>> >>>-- >>>Tim Gruene >>>Institut fuer anorganische Chemie >>>Tammannstr. 4 >>>D-37077 Goettingen >>> >>>GPG Key ID = A46BEE1A >>> >>> >>>On Fri, 2 Dec 2005, [gb2312] 李昂 wrote: >>> >>> >>> >>>>I used refmac5 to refine a structure out from MR. But it seems >>>>the geometric parameters of the output model are really bad, >>>>only 50 residues are in core region of Ramachandrans plot, all >>>>output from procheck are not good, Rfree value of the model is about >>>>45. Can someone give any suggestions? Is there any special >>>>parameters needed when the refinement is on some initial moder? >>>>I didn't met such problem on models that are almost correct. >>>> >>>>Any help are thankful! >>>> >>>>Yuhe Liang >>>> >>>> >>>> >>>> >>>> >>>> > >= = = = = = = = = = = = = = = = = = = = > > > > > > > > > > > >
begin:vcard fn:Eleanor Dodson n:Dodson;Eleanor email;internet:[EMAIL PROTECTED] tel;work:+44 (0) 1904 328259 tel;fax:+44 (0) 1904 328266 tel;home:+44 (0) 1904 424449 version:2.1 end:vcard
