On Wed, Oct 26, 2011 at 5:00 AM, Jan-Philip Gehrcke <jgehrcke.googlemail.com
> wrote:
> Thanks. I've a few comments left:
>
> 1) In the official AmberTools manual
> (http://ambermd.org/doc11/AmberTools.pdf), I believe it is not stated at
> all that for decomposition, sander 11 is required.
>
May not be. I'll add a note.
>
> 2) However, I've once saved a manual "MMPBSA_Python_Manual.pdf" (I
> cannot find from where I've downloaded it). There it is stated "&decomp
> namelist variables (available only for Amber 11)". This would have made
> me suspicious, but "must not be used with an Amber version smaller than
> 11" is unambiguous. Not "available" sounds like an error should be
> printed if I still try to use it.
>
Indeed, the version that Manual applied to did throw an error, as long as
nobody renamed their extracted directory from amber10. :) Some of the
things fixed since that version are that MMPBSA.py is well-versioned now (so
there's no hunting around to see what version of the code you're using), and
it's printed in the output file.
> 3) During our discussion in this thread, you, Jason, have mentioned "the
> gbsa=2 code has known issues (at least 1 bug)". Is this worrying for
> me/us, even if I will use sander 11?
>
Meh, I don't really think so. The deviations are small, and it's multiplied
by a small surface tension term, anyway. Just be wary of large differences
in the surface tension terms between complex/receptor/ligand. (You could
also set surften==0 and ignore it altogether).
> 4) Doesn't sander have something comparable to a "--version" switch?
> Anything that could be used to certainly identify the code version? A
> check based on this and a meaningful error message would be a clean
> solution.
>
Kind of. The version is printed in the very top of the mdout file. This
could be used -- it had never occurred to me to check this. However, I
couldn't figure it out before sander was run. The problem is that you were
clever enough to hunt through the code and allow MMPBSA.py to pick out
binaries that were not in $AMBERHOME/bin (which is only amber11 for
AmberTools 1.5), so unless this change was made, sander10 wouldn't be
available (unless they linked in sander10 into the amber11/bin).
> 5) We've "already" ordered Amber 11 at beginning of October and got a
> mail on Monday that we'll receive download instructions within a week.
> Now, I suddenly cannot wait to run MMPBSA.py using sander 11 :-) What's
> happening within that week?
>
Not quite sure I follow... (Make sure you run the AT15_Amber11.py script
according to the directions when you build Amber11 alongside AmberTools
1.5).
All the best,
Jason
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Oct 26 2011 - 08:00:02 PDT
