Hello,
Is it possible to alter the weight of homology-derived restraints, in the same way you can scale other contributions to the energy function (with 'environ.schedule_scale')? I would like to globally down-weight homology-derived restraints to place more emphasis on geometry/electrostatics etc.
Any advice would be much appreciated, as was the information on SAXS restraints.
Cheers,
Nick
---------------------- Dr N M Burton Department of Biochemistry University of Bristol BS8 1TD, UK +44 117 3312149 nick.burton@bristol.ac.uk
On 07/08/2009 02:35 AM, Nick Burton wrote: > Is it possible to alter the weight of homology-derived restraints, in > the same way you can scale other contributions to the energy function > (with 'environ.schedule_scale')? I would like to globally down-weight > homology-derived restraints to place more emphasis on > geometry/electrostatics etc.
Yes, certainly. The homology-derived restraints are not all placed in the same restraint group though, so you'll have to down-weight multiple groups (on the other hand, this gives you a little more flexibility, if you only wanted to down-weight the distances or dihedrals for example). Look in the log file from a comparative modeling run for the "Summary of the restraint violations", particularly the RESTRAINT_GROUP, NUM and S_i columns, and http://salilab.org/modeller/9v7/manual/node245.html for a list of all the restraint groups.
Since there are many homology-derived restraints, it may make more sense to simply up-weight the restraints you want to put more emphasis on - the weights can be greater than 1.
BTW, by default Modeller does not consider electrostatics, since the assumption is made that any electrostatic interactions are already represented by the template conformation, and without solvation the interactions are unlikely to be correct even if you do turn them on.
Ben Webb, Modeller Caretaker
OK, I didn't realise that restraints listed in those locations were a break-down of the homology-derived restraints. Is there anywhere which details which are homology-derived and which aren't (eg are the bond length/angle potentials homology-derived, or based on dictionary values)?
Thanks also for the advice on electrostatics.
Cheers,
Nick
--On 08 July 2009 11:00 -0700 Modeller Caretaker modeller-care@salilab.org wrote:
> On 07/08/2009 02:35 AM, Nick Burton wrote: >> Is it possible to alter the weight of homology-derived restraints, in >> the same way you can scale other contributions to the energy function >> (with 'environ.schedule_scale')? I would like to globally down-weight >> homology-derived restraints to place more emphasis on >> geometry/electrostatics etc. > > Yes, certainly. The homology-derived restraints are not all placed in the > same restraint group though, so you'll have to down-weight multiple > groups (on the other hand, this gives you a little more flexibility, if > you only wanted to down-weight the distances or dihedrals for example). > Look in the log file from a comparative modeling run for the "Summary of > the restraint violations", particularly the RESTRAINT_GROUP, NUM and S_i > columns, and http://salilab.org/modeller/9v7/manual/node245.html for a > list of all the restraint groups. > > Since there are many homology-derived restraints, it may make more sense > to simply up-weight the restraints you want to put more emphasis on - the > weights can be greater than 1. > > BTW, by default Modeller does not consider electrostatics, since the > assumption is made that any electrostatic interactions are already > represented by the template conformation, and without solvation the > interactions are unlikely to be correct even if you do turn them on. > > Ben Webb, Modeller Caretaker > -- > modeller-care@salilab.org http://www.salilab.org/modeller/ > Modeller mail list: http://salilab.org/mailman/listinfo/modeller_usage
---------------------- Dr N M Burton Department of Biochemistry University of Bristol BS8 1TD, UK +44 117 3312149 nick.burton@bristol.ac.uk
Nick Burton wrote: > OK, I didn't realise that restraints listed in those locations were a > break-down of the homology-derived restraints. Is there anywhere which > details which are homology-derived and which aren't (eg are the bond > length/angle potentials homology-derived, or based on dictionary values)?
The best source for this is the 1993 Modeller paper, but there is also a brief description at http://salilab.org/modeller/9v7/manual/node196.html
Bond lengths, angles and some dihedrals come directly from the CHARMM forcefield, so are not homology-derived. But phi/psi and sidechain dihedrals take the template residue type into account, and so are homology-derived.
Ben Webb, Modeller Caretaker
participants (2)
-
Modeller Caretaker -
Nick Burton