Hi Michael,
Thanks. It seems like you solve the problem partially by
deleting some of the unhappy restraints. I usually get no NOE
violations and several CDIH violations. I do not know whether there
is a way to adjust the Nonbond parameters to solve the problems.
Anyhow, I will try to delete some of the CDIHs and figure out
the problems.
Thanks for sharing your experience.
Lei
On May 12, 2007, at 9:18 AM, Michael E Hodsdon wrote:
>
> Lei,
>
> Hi. I am not sure if I really solved the vdw problem. I did manage
> to get an ensemble of structures with essentially zero violations
> of my NOE, CDIH, HBDA and RDC restraints. The RAMA potential is
> very low (very negative), so I think that worked well, but I still
> get around 10-15 vdw violations per structure. When I run the
> structure through the PDB validation server, it found hundreds of
> "close contacts". They are all very tiny violations and the rest of
> the geometry checks out very nicely (almost too good for the chi1
> and chi2 angles!).
>
> I think that part of the issue is that there are tiny differences
> between everyone's vdw radii definitions (i.e. xplor vs. CYANA vs.
> PROCHECK, etc.). I am not sure how important these tiny differences
> are.
>
> I can tell you how I solved the restraint violation problem. I just
> followed Marius' advice and got rid of my stereospecific
> restraints, loosened my NOE restraints, and went through a few
> rounds of deleting "bad" restraints. I made a bunch of mistakes
> along the way, but basically this was the protocol. One big thing I
> learned that it is a BAD idea to "loosen" RDC restraints by
> increasing the error. I beleive it just increases the relative
> weight of the RDC that is causing trouble. I spent a long time
> trying to get convergence doing this and it was just getting worse
> and worse. I then went back to the "beginning" and tried simply
> deleting the 4 or 5 RDCs that were giving me trouble and things
> converged very quickly after that.
>
> Good luck and let me know how it goes.
>
> Mike
>
> P.S. If you could forward this to the Xplor-NIH mailing list or if
> you reply then just reply to the mailing list, I would greatly
> appreciate it. I would like to see if Charles or Marius (or anyone
> else) has any comments on this issue. For some reason, the only way
> I can send something to mailing list is to use a web-based email
> program (instead of my preferred Thunderbird) and it is a real hassle.
>
> Lei Shi wrote:
>> Hi Michael,
>> I was just reading the threads of xplor mailing list.
>> http://nmr.cit.nih.gov/pipermail/xplor-nih/2007-March/
>> 000573.html
>> I have the same problem you have with the large VDW
>> violations. I am using the default parameters in the refine.py.
>> How do you solve the problem? Did you do it by changing different
>> parameters?
>> Many thanks,
>> Lei
>
> --
>
> Michael E Hodsdon, MD, PhD
> Assistant Professor of Laboratory Medicine and Pharmacology
> Yale University School of Medicine
> email: michael.hodsdon at yale.edu
> http://info.med.yale.edu/labmed/faculty/hodsdonm.html
>
> 203.737.2674 (office) 203.737.2684 (lab)
> 203.915.0780 (mobile) 203.688.8704 (fax)
>
> Administrative Assistant: Vilma Moreno
> phone: 203.737.5934; email: vilma.moreno at yale.edu
>
> U.S. Postal Address:
> P.O. Box 208035, Laboratory Medicine
> Yale University School of Medicine
> 333 Cedar Street
> New Haven, CT 06520-8035
>
> Shipping Address:
> CB462b, Laboratory Medicine
> 200 South Frontage Road
> New Haven, CT 06520
