The problem you're facing has nothing to do with the NOEs. Rather, it's that you're not doing a particularly good job at minimizing the initial covalent geometry. You begin with setting the coords of every atom to zero (confusingly, in two different places). You then set the flags to include every covalent term and a bunch of experimental ones. You then set a very short (50 step) minimization. Not surprisingly, the minimizer gives up quickly (note the "line search abandoned" note in the output). This leaves grossly violated bond restraints, which require very short MD timesteps to simulate accurately, because of their extremely high force constant. Since you're using a 5fs timestep, you're asking for the MD to blow up on you. Sometimes, you might survive the MD (as appears to be the case with the center averaging), but this is just luck.
The minimizer gets stuck easily in local minima. Therefore, if you want to minimize coords with very bad covalent geometry, you need to take it in stages, starting from the simplest energy hypersurface and progressing to more complex ones. Thus, something like vector do (x = 0) (all) vector do (y = 0) (all) vector do (z = 0) (all) flags exclude * include bond end mini powell nstep 100 nprint 10 end flags include angle end mini powell nstep 500 nprint 10 end flags include impr end mini powell nstep 1000 nprint 10 end will get you a lot closer. It won't work every time, but it'll reduce the number of covalent geometry violations to a level that (probably) won't break the MD. If you drop your MD timesteps to 1fs, you'll increase the odds further. --John Kuszewski On Wednesday, November 19, 2003, at 06:09 PM, Evgeny A. Fadeev wrote: > Charles, > here is some small input file with one test noe record and its result. > > it works fine whenever i use 'cent' averaging > or not use noe at all, but fails whenever i use other option. > > location of toppar file will need to be adjusted > to run it > > i have xplor version 3.85, i am not sure how versions are named, it > seems that currently there is > 2.9~ available, could version 3.85 be the source of trouble? > > thank you. > Evgeny > > > > ----- Original Message ----- > From: [email protected] > Date: Wednesday, November 19, 2003 4:04 pm > Subject: Re: [Xplor-nih] SUM averaging option > >> -----BEGIN PGP SIGNED MESSAGE----- >> Hash: SHA1 >> >> >> Hi-- >> >> >>> >>> the problem is that whenever i change >>> >>> noe >>> ... >>> averaging * cent >>> .. >>> end >>> >>> to >>> ... >>> >>> averaging * SUM >>> >>> in the original sa.inp file >>> >>> calculation breaks and program complains that it >>> does not know coordinates of atoms that happen to participate in >>> ambigous noe contacts >>> >>> R-6 option also failed >>> >>> I wonder if anybody knows how to get around this? >>> >> >> If you could send a bit of output with any error messages, it >> would be >> helpful. sum averaging is the most commonly used, so it should work >> just fine. >> >> regards-- >> Charles >> -----BEGIN PGP SIGNATURE----- >> Version: GnuPG v1.2.1 (GNU/Linux) >> Comment: Processed by Mailcrypt 3.5.8 <" >> target="l">http://mailcrypt.sourceforge.net/> >> iD8DBQE/u9rSPK2zrJwS/lYRAmSUAJ9PnTpSvQZpoRds7lXvNQdGZ9f8awCePFeU >> B+4ipYLwdQAAl/YZy+uE8rQ= >> =ImhS >> -----END PGP SIGNATURE----- >> > <butane.out><butane.inp>_______________________________________________ > Xplor-nih mailing list > [email protected] > http://dcb.cit.nih.gov/mailman/listinfo/xplor-nih
