-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Hello Ramon-- > > I'm trying to reproduce a structure of a nucleic acid (RNA in this > case). > I built an RNA helix using W. Olson's 3X3DNA, calculated the distances > and implemented them as restraints in xplor. > (PDB validation gives no errors on angles and bond-lengths, and NOE > backcalculation gives no violation for a severely overdetermined > system). > > When I run a POWEL or short verlet dynamics the structure runs away from > the perfect conformation (only potentials included are: noe bonds vdw > angles, impropers) > I see a huge increase in energy of the bonds and angles, while the > structure is initially in it's energy minimum. Two things to look at first: 1) what are the reported violations of the initial structure (obtained using print threshold=0.95 bond, etc)? 2) how much does the minimized structure differ from the initial one? If it differs by much, I would first suspect that the nonbonded term is misconfigured: have you set rcon, etc? If vdw and noe terms are properly configured, the structures should move very little upon minimization. > > I checked the topology and parameter files: > topallhdg.dna and parallhdg.dna Please use nucleic.top and nucleic.par instead. These contain multiple fixes/improvements over the files you used. We do plan to ship updated toppar files in the next release of Xplor-NIH- the fixes primarily involve making Xplor-NIH atom naming more IUPAC-compliant. > > They contain the wrong values for bondlengths and angles. > Meaning, they do not match Olson's model (which is correct according to > pdb validation-software from rcbs.org, and also according to the > references: > JACS (1996), vol.118 iss. 3, pp. 509-518 > JACS (1996), vol.118 iss. 3, pp. 519-529 > > An example: O4'-C4' and O4'-C1' have the same length, whereas they > should be 1.453 and 1.414 respectively. The current Xplor-NIH value for these two lengths is 1.43- about the average value, and within the standard 0.05 angstom threshold which we typically use for bond lengths- so the cited values match that in nucleic.par to within expected accuracy. I would say that the values in these papers are more precise for a particular conformation, but do note that these parameters vary with sugar conformation. Looking at the histograms of the two bond lengths in the second paper shows that they overlap significantly. Anyway, I don't think this is the your problem at the moment. best regards-- Charles -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux) Comment: Processed by Mailcrypt 3.5.8 <http://mailcrypt.sourceforge.net/> iD8DBQFFy4HQPK2zrJwS/lYRAuJkAKCCpskWJkxE6+hJQ/gYCNvmtdBtoACaAzhj 7KgGTyQIWs+rnT5PUui65Vk= =YVV8 -----END PGP SIGNATURE-----
