Eric Martz wrote:
A few days ago, Miguel asked whether altLoc's were correlated (e.g.
all A's go with all other A's as a "model") or not (each A vs. B
altLoc is independent of the other A's and B's).
Scott Garman, a crystallographer, kindly sent some clarification which
I will paraphrase as follows. Experimentally, altLoc's are assigned
when the electron density supports two distinct positions for the same
atoms (in different copies of the molecule in the crystal). (In
unusual cases there could be evidence for three or more positions, but
often the electron density would get too blurred to be sure beyond two.)
If only side-chain atoms, (CB = carbon beta, CG = carbon gamma, NH =
nitrogen eta, etc.) have altLoc's, each side chain's A altLoc
positions are independent (uncorrelated) with the A altLoc's of other
side chains.
If both side-chain and main chain (CA = carbon alpha, N = main chain
nitrogen, C = main chain carbon) atoms have altLoc for a consecutive
run of amino acids, then they are alternate positions for the entire
run, so all the A altLoc's are one local conformation, and all the B
altLoc's are another.
I say "local" because farther down the chain could be another run of
altLoc's A and B, but the second run of A need not go with the first.
The first run could be in conformation altLoc A, while the second is
in B in the same molecule.
I totally agree, the combinatorial view is true for the experimental
structures in the PDB.
So the bottom line is that altLoc's can be uncorrelated or locally
correlated in a consecutive run, but still uncorrelated between
different runs.
I don't think Jmol needs to have any intelligence built-in about this,
beyond that the implementation should allow for the absence of
correlation between different A's or runs of A's -- it certainly
should not assume that all altLoc A's are "one model".
but because this my lead to problems with intuitive model naming, there
may be three strategies:
1. don't be intuitive, just count
2. despite the experimental truth (where no one knows whether 1%A
will coexist at the same time with 13%A or not) in most cases where
animation is desired, these are model assemblies like the rotamer
distributions of Joost Van Durme, and the author has its hands on the
naming, e.g. using 1:A%A 1:A%B and 13:A%C 13:A%D
3. use the full descriptor to address the whole model
/<MODEL>[SER]1:A%A[*]13:A%A[*]<rest of residue selector>%<altLoc>
or a select string
/<MODEL>&(1%A,13%A & *:A)
though my main aim had been to get the RasMolScript *;A altLoc selector
implemented and I started argue for 2, now I think Jmol should do it
right and should offer both 3. and 1. possibly by an analog command
like the RasMolScript renumber command does for sequence numbers e.g. as
a set command
set model renumber
/1
/2
/3
/4
set model normal
/1&[*]*%A
/1&((*%A,1:A%B) & not 1:A%A)
/1&((*%A,13:A%B) & not 13:A%A)
/1&*%B
Regards, Jan
-------------------------------------------------------
This SF.Net email is sponsored by: NEC IT Guy Games. How far can you shotput
a projector? How fast can you ride your desk chair down the office luge track?
If you want to score the big prize, get to know the little guy.
Play to win an NEC 61" plasma display: http://www.necitguy.com/?r=20
_______________________________________________
Jmol-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/jmol-users
