[email protected] wrote:
Paul Emsley wrote:
wtempel wrote:
Anyway, suppose I have "split" a residue, say lysyl, to model
alternate conformations of its side chain. Suppose further that I
would like to remove NZ from conformer B. Upon doing so, the current
COOT implementation will remove the atom. Good. It will also remove
the confID from what was conformer A's NZ atom and re-set its
occupancy to 1. On the latter part, intuition tells me that instead,
conformer A's atoms should be unaffected, namely keep their confID and
sub-unity occupancy.
My questions:
1) What behavior is expected by other COOT users in comparable
situations?
Would the answer be different if a C or an O had been deleted? I'd
rather ask: is Coot doing the right thing in such a case? I suspect
that Garib or Pavel may know.
My problem is that I agree with the original poster that the remaining
NZ atom should continue with a confID of A, but I also agree with Paul
that the deleted C atom should be changed to a single conformation.
However, if the residue that the C atom is linked to is also split
I would feel differently. In my favorite structure there is a stretch
of five residues where the main chain has two conformation. If I deleted
a C atom from the B conformer in the middle it would not be my intention
to have only a single conformation for that atom.
On investigation, the PDB documentation has this to say: "If an atom is
provided in more than one position, then a non-blank alternate location
indicator must be used. Within a residue, all atoms of a given
conformation are assigned the same alternate position indicator. "
It does not say that if there is only one position of an atom in a
residue then that atom must have a blank alternative location indicator.
I presume that if they had meant that to be so, then they would have
said it.
How about this suggestion? If the deleted atom is bonded to an atom
with a single conformation then the other conformer becomes a single
conformation. If the deleted atom makes bonds only with atoms that
also have alternative conformations then the remaining atom retains
its partial occupancy. This would keep the poster happy, keep Paul
happy with his C atom when the neighbor has a single conformation,
and will keep me happy when there is a stretch of main chain disorder.
So it seem the answers to Wolfram Tempel's questions are:
1) Don't fiddle with the alt confs of the other position (unless it is
bonded to any atom with a blank alt conf indicator - or perhaps only
bonded to atoms with blank alt conf indicators (not sure)).
2) Yes.
A problem with this scheme occurs with the lysine mentioned before.
If the user decides to delete the entire B conformation one atom at a
time the result will differ depending on the end of the side chain
they start deleting atoms from.
Implementation issue of which I must take note.
