[phenixbb] Distance constraint to atom in symmetry copy in phenix.refine
Pavel Afonine
pafonine at lbl.gov
Fri Mar 22 19:38:48 PDT 2019
Hi Julian,
I'm glad that was helpful! Please let me know should you have any more
questions or need any assistance! You are correct, there is no plane
restraints that are symmetry aware -- this must very tricky to implement!
All the best,
Pavel
On 3/22/19 15:58, Julian Esselborn wrote:
> Thanks a lot Pavel! This was exactly what I was looking for and now I
> realized I should have seen that myself.
>
> Worked for one of my problems, where I just needed a distance
> restraint. Unfortunately I also had one, where I would have needed a
> planarity restraint and there is no symmetry option for those I think.
> So I ended up using the alternative position workaround for that one.
>
> Am 04.03.2019 um 17:53 schrieb Pavel Afonine:
>> Hi Julian,
>>
>> perhaps you can approach this by defining custom bonds between
>> symmetry copies, like this:
>>
>> refinement.geometry_restraints.edits {
>> bond {
>> atom_selection_1 = chain A and resseq 123 and name N
>> atom_selection_2 = chain B and resseq 321 and name OD1
>> symmetry_operation = -x-1/2,y-1/2,-z+1/2
>> distance_ideal = 2.1
>> sigma = 0.02
>> }
>> }
>>
>> Pavel
>>
>> On 3/1/19 17:22, Julian Esselborn wrote:
>>> Dear community,
>>> we have a somewhat complicated problem to which I don't seem to find
>>> a solution.
>>>
>>> We have a structure, which has a number of 3-fold and 2-fold
>>> symmetry axes in the final assembly structure. The 3-fold axes are
>>> hold together by metal atoms on the axis.
>>> However, we have three cases of these axes:
>>> 1. Symmetry axis falls onto the crystallographic symmetry axis. We
>>> can deal with this; setting metal to 0.33 occupancy and setting
>>> metal-protein distance constraints. This is a proper symmetry axis.
>>> 2. Symmetry axis doesn't fall onto a crystallographic symmetry axis,
>>> but all three monomers coming together are within the same symmetry
>>> copy of the ASU. Even easier, metal stays at occ=1 and we just set
>>> constraints to chain A, chain B, chain C.
>>> 3. The really challenging case, where the axis doesn't fall onto the
>>> symmetry axis, but the three monomers coming together are in
>>> different symmetry copies of the ASU.
>>>
>>> Cases (2) and (3) are pseudo-symmetric in the crystallographic sense.
>>>
>>> Usually a bit of intelligent moving around of the monomers to their
>>> crystallographic symmetry positions should push all monomers of case
>>> (3) into neighboring positions within the same ASU ending up as case
>>> (2); problem solved.
>>>
>>> BUT: In our structure we have too many 3fold axes, such that there
>>> will always be one of them ending up as case (3). E.g. the three
>>> monomers are chains A, B, C, but they do not end up neighboring in
>>> the ASU. Rather the axis is formed by A, B' and C'' (with ' denoting
>>> different symmetry copies of the ASU). We could assign the metal to
>>> chain A with occ=1 and no metal in B and C. However, we would need
>>> to set a distance constraint from the metal to it's ligands in
>>> protein monomer B and protein monomer C. But it is only the ligand
>>> in B' and C'', which are actually close to the metal in A. The
>>> ligands in B and C are at the other end of the ASU.
>>> Is there a way to set a distance constraint such that it measures
>>> the distance to a crystallographic symmetry copy?
>>>
>>> A different idea was to just assign an alternative position alt A,
>>> alt B and alt C to the metal, with A being close to the ligand in
>>> monomer A, B close to monomer B and C close to monomer C. That way
>>> we could make constraints. But we would need to cross fingers that
>>> the three metal atoms actually end up in the same spot once applying
>>> the crystallographic symmetry (and remember, that 3-fold axis is not
>>> constructed by applying a 3fold rotational symmetry around that
>>> axis; rather an actual crystallographic symmetry somwhere else
>>> brings them together; means the ligands with their metal atoms can
>>> actually move independently).
>>>
>>> I'm a bit at a loss how to deal with this and would appreciate input.
>>>
>>> Thanks a lot in advance!
>>>
>>> all the best
>>>
>>>
>>> Julian
>>>
>>>
>>>
>>> ----
>>> Julian Esselborn
>>> Postdoctoral Researcher
>>> Tezcan group
>>> University of California, San Diego
>>
>
More information about the phenixbb
mailing list