OK, I updated ccp4 to 6.2.0 (and refmac to 5.6.0117) and now the angles coming out close to the ideal values. Also, I see now the the "MODRES" in the pdb header from before was tell future refmac runs to rename those residues - which confused the newer refmac version as it thought the DT (renamed Td) were somehow "DY". By just deleting these MODRES lines it was fine.
On Sep 14, 2011, at 7:10 PM, Gregory Bowman wrote: > I'm running into some geometry problems with my DNA model after refinement > with refmac (version 5.5.0109), and would appreciate any feedback. > > The problem is that the angles for many of the glycosidic bonds are 2 to 4 > degrees off of the ideal values, and so are several standard deviations > outside what's expected. Our data (from MR) is ~2 Å, and the density is well > defined for the DNA. I was thinking that perhaps refmac was not recognizing > my DNA and so just taking the poor geometry from one model and enforcing it > for the next, but in the log file it does not indicate that a new molecule > has been found, and I do not feed in a cif file from previous runs. Another > indication that refmac recognizes the DNA bases is that it says it is > renaming them. The DNA bases are named DA, DC, DT, DG, and the header of the > output pdb file says: > > MODRES DC B 1 Cd RENAME > MODRES DC B 2 Cd RENAME > MODRES DA B 3 Ad RENAME > MODRES DT B 4 Td RENAME > (etc...) > > In the output, strangely (to me), it is actually not renaming them. That is, > it is keeping the "DC" etc names. > > In coot, I'm measuring the glycosidic bond (O4'-C1'-N9) to be 104.6, whereas > in AD.cif, it is listed as 108.4. > > The bond lengths are fine, and there are no distortions of the protein model. > > Thanks, > Greg > -- Department of Biophysics Johns Hopkins University 302 Jenkins Hall 3400 N. Charles St. Baltimore, MD 21218 Phone: (410) 516-7850 (office) Phone: (410) 516-3476 (lab) Fax: (410) 516-4118 [email protected]
