[phenixbb] geometrical restraints between symmetry related chains

[Jorge Iulek]

Dear all,

	I am working currently on a structure that should be "peptide bond" 
between chains, but the "second" part of the chain is in a symmetry 
related peer. Numbering is continuous and there are justifications to 
why the chains are separately named with different chain ID's. By the 
way, this "connection" is "full of glycines" so a kind of flexible, but 
a continuous stretch can be seen at "reasonable" density (and despite 
low resolution).
	To better explain, say I have chain C residue x to be bound to chain K, 
at symm operation -x-1,y-1/2,-z-1/2, residue x + 1.
	I would prefer to really refine the presence of these "connecting" 
glycines, of course with due geometrical restraints imposed (precious at 
this low resolution).
	phenix.refine offers:

refinement.geometry_restraints.edits

at

bond

with the possibility to indicate the

symmetry_operation

	for the "second atom" within the bond. But this "symmetry_operation" 
indication is not available (?) for the other geometry_restraints edits, 
like  angle, dihedral, etc. (eventually, planarity might also make 
easier). Do I miss anything here?
	Currently, I see 3 possibilities:
a) somehow (how?) I can add angle, dihedral restraints in involving 
chains in symmetry related operations
b) I bring the symmetry related chain to its non "symmetry related" 
position and change chains names/residues ID, etc., so I go into this 
go/return of naming at each step of the refinement work/analysis
c) I add "extra" ( + - like alternate conformation residues) residues to 
each end of the connection, such that these extra residues will receive 
due conventional geometrical restraints (as phenix.refine does for 
contiguous residues in a chain) but I force them to be on the same 
position (and probably declare 0.5 occupancy for these overlying residues).
	"a" should be the easiest, if available somehow; "b" would be 
cumbersome during the refinement/analysis process; "c" seems to be the 
most complex but maybe the best if it works. Nevertheless, someone might 
have an easier suggestion.
	Thanks,

Jorge