Re: [modeller_usage] Restrain Uses in MODELLER v.91.6

[Sushil Mishra <>]

On Tue, Oct 4, 2016 at 3:43 AM, Modeller Caretaker
<> wrote:
> On 10/3/16 1:47 AM, Sushil Mishra wrote:
>>
>> I am trying to model a protein with certain distance restraints and
>> disulphide bonds. I wonder why restraint does not work when I set
>> disulphide bonds.
>
>
> What do you mean by "does not work"? Your script looks fine to me.

I mean, in the generated models those restrained atoms (CA of 1 and
169) are far away (>20A) while distance should be somewhere between
within 8A. The CA atom of residues 1 and 169 are close enough in the
model when function special_patches is commented. However, I am
missing a S-S bond without this function. I do not understand why
those two CA atoms are >20 in the model when there is restrain set
between them:

rsr.add(forms.gaussian(group=physical.xy_distance,
                               feature=features.distance(at['CA:1'],
                                                         at['CA:269']),
                               mean=4.0, stdev=4.0))


>> What can be the
>> reason that it does not work with S-S restraints. Moreover, Is there a
>> way to control force constant for harmonic restraints in such cases ?
>
>
> Technically, you are not adding an S-S restraint - you are telling Modeller
> that the two disulfides are bonded, so when it turns on the CHARMM
> forcefield you'll get a term in the scoring function that corresponds to the
> S-S bond. So if you want to adjust that force constant (I can't imagine why
> you'd want to) the way to do it would be to edit the CHARMM parameter file,
> modlib/par.lib (in the BOND section, which is arranged by CHARMM atom type;
> the two S atoms are type SM, so you'd be looking around line 291 in Modeller
> 9.17).

I understand it now. I guessed it incorrectly. I thought MODELLER may
be using this information to restraint both Cys residues during the
modeling so that they are close enough to form S-S bond.

Thanks,
SM


>
>         Ben Webb, Modeller Caretaker
> --
>              https://salilab.org/modeller/
> Modeller mail list: https://salilab.org/mailman/listinfo/modeller_usage