Thanks for catching this, Ye Mei,
There is not in fact a typo in that tutorial, but the reasoning is very
subtle. It's rather arbitrary what you call "solute atoms" as I am
exploiting (relying on?) the fact that AMBER tleap places the protein
first and all water last. You could in fact run all three commands with
-P 4143 and probably be fine. Note that the -P option is really
meaningless, at least for the first recursive iteration of the third
command, because all it does is distinguish whether atoms shall be treated
with the -RP cutoff or the -RW cutoff, and unless the third command fails
to add 179 waters on the first run-through both -RP and -RW are equal (the
-V 1 option tells AddToBox to call itself recursively with iteratively
smaller values of -RW if it cannot add enough of whatever molecule, water
in this case).
So, as for your question, the oddity is really in the second command. I
chose to treat the added acetate molecules as part of the "solute" in the
second command because I don't mind if some ammonium is added 3.0A from
the acetate (it may well exist that way in solution), but I do not want
ammonium added very near other ammonium to make sure it's all spread out
(at least, initially). I may have also had a hard time getting that
second command to work with -P 4143 ("treat the acetate as part of the
solvent, and make sure ammonium cannot be added with 6.0A of acetate"), I
don't recall. Finally, you may find that running the third command with
-P 4207 would be a tall order, as that makes a lot of the volume
off-limits to water molecules. Very bad if water is placed too close to
protein atoms, agreed, but not really so bad if it's placed close to
ammonium or acetate--the water has only one vdW site, so it effectively
cannot become "entangled" with the acetate, but imagine if you were adding
lots of something big, like hexane--allowing hexanes to be places such
that two atoms could come within 1.5, 1.35, 1.21A of one another (as would
happen if you were trying to pack the cell and allowing AddToBox to
iteratively relax the criteria for adding new molecules) might let you get
two hexanes tangled up with one another, which would eventually manifest
itself as a vlimit error in production runs.
The big picture, as I hope I said in the tutorial, is "big and rare
molecules first, ubiqutious and small molecules last." Understand, there's
some ambiguity there that the tutorial doesn't run into but your question
does. What if you have to add some molecules that are big and some others
that are rare? What if your crystallization solution is 50% hexane with
some trace ions? Which goes first? Well, the answer is hexane first, and
give it a reasonable -RP and a good initial -RW to make sure the hexanes
don't get tangled up with one another. Then, add your ions, making hexane
part of the "solute" and placing a big -RW to spread the ions out, and
finally add the water, probably making only the protein the solute, to let
water fill any little gaps between hexanes. I hope that helps.
Dave
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Oct 15 2010 - 09:00:05 PDT
