Hi Tim,

altloc canbe used as a way to manipulate occupancy refinement and non-bonded interactions. The rule is:

- see each other via non-bonded:

altloc " " <> altloc " "
altloc "A" <> altloc " "
altloc "A" <> altloc "A"

- do not see each other via non-bonded:

altloc "A" <> altloc "B"

If you have a molecule on a symmetry element then there is a special (additional) way to turn the symmetry non-bonded interactions off.

So, as long as you can manage to assign altlocs such that 1) you achieve desired occupancy refinement behavior and 2) desired non-bonded interactions, you are fine.

Conveying this to PDB is a different story, I think. There is a number of meaningful scenarios where there is no obvious way of documenting it using altlocs.

I'm not aware of any consensus here (which obviously doesn't mean it doesn't exists!).

All the best,
Pavel

On 6/14/12 8:53 AM, Timothy Springer wrote:

We have a 1.4 Å structure with two pseudo-symmetry related A and B chains that form a super b-sheet. An 8-residue loop has two quite different alternate, backbone conformations. When we refine the occupancy of these loops, they come out to 0.51 and 0.49, consistent with them being mutually incompatible (i.e., forced to be 0.50 and 0.50 in the ideal situation). Indeed, clashes would exist between the A chain A alternate and the B chain A alternate, but when these are A and B alternates, there is no clash. Since these are in separate chains, Phenix has no problem with an A and A clash. There is a second loop, non-contiguous in sequence, that also has clashes with one of these. We refine each of the two clashing loops as the A conformat ion. Since these are separate loops, Phenix does not seem to care and seems to treat the A conformation of one loop as if it is compatible with the B conformation of the other loop.

This is great that Phenix is so forgiving, but my concern is that shouldn't by convention we reassign the conformations so that the A conformations are all compatible within the same monomer? Is there some formal convention on this, and will PDB ask us to reassign these?

In a closely related 1.7 Å lattice, we have a single molecule per asymmetric unit, and now the relationship is between two symmetry-related molecules. The two conformations of loops are similar as in the 1.4 Å structure. Here, we have to keep the same alternate conformation name, and thus have a clash between A and A loops 1 but not A and B loops 1. Again there is a clash between A conformation of loop 1 and A conformation of loop 2, and we could make this go away by switching A and B in loop 2.

I find it to be less confusing if we switch the conformation names in loop 2. What is the consensus view?

Tim