From chemistry-request@ccl.net Wed Jun 19 00:48:24 1991
Date: Tue, 18 Jun 91 20:39:47 PDT
From: ross@zeno.mmwb.ucsf.EDU (Bill Ross)
To: apa@ccadfa.cc.adfa.oz.au
Subject: Re:  Free-energy perturbation calcs of zinc<->peptide binding
Status: R

	I have parameterised the zinc-ion in AMBER 3.0A as an 'H-bonding' type 
	of atom ie. it has 10-12 well-depth and equilibrium bond distances with
	all relevent donor atoms on the peptide (peptide C=O, carboxylate-O,
	amine H2N's etc).  

Why do you see zinc as hbonding? If it were negatively charged there might
be some basis. 6-12 would be the way to go, I believe. 

These calculations are sensitive to the nonbonded potential you choose, 
so I suggest serious thought on how you determine 6-12 parameters (or 
10-12 if there's a reason for it). See e.g. J. Aqvist, J. Phys. Chem. 
1990, 94, 8021-8024, and remember that ion-water parameters might be 
different from ion-peptide ones when water is modelled as a big sphere 
with a few charges in it.

	In addition, it has a point-charge calculated by fitting
	the electrostatic potential with MOPAC.  

What charge results? Was this for the ion alone? Why not use 2+ ?

	Am I right in thinking that I need to do 3 molecular dynamics 
	simulations:

	(i) using MD to equilibrate the previously minimised structure to 298K -
	this may take 10ps or so with 1-2fs steps. Save the 'equilibrated' 
	velocities and coords.

You might want to warm carefully and measure the geometry around the
ion especially to get an idea of the time period of the motions and
as a guide to when equilibration is satisfactory.

	(ii) start the GIBBS slow-growth using the equilibrated velocities and 
	coords with IELPER=1 and vary LAMBDA from 1->0 over say, 20ps.  

Then run backward to check your hysteresis before continuing. You
could even run forward to say .8 & then back to see how long your
run should be (longer run => less hysteresis). Since the electrostatic
effect is strong and long range, you may need 50-200 ps each way. You
may want to break it up into several runs so that finer granularity 
is used at higher charge density.

A safety measure might be to equilibrate some more at 0 charge before
continuing.. (requires a special parm topology file for the 0 charge
or running a window perturbation with only a long equilibration 1st window).

	(iii) start GIBBS again using the equilibrated velocities and coords 
	with IELPER=-1 and again vary LAMBDA from 1->0 over the same interval.

Then run the gibbs(iii) back from small, non-0 lambda to check hysteresis.
(Running backward from 0 lambda gives a ghost that can 'appear' on top of 
another atom.) My guess is that this perturbation can be much shorter
than the charge one.

	How do I decide if 20ps was long enough? 
	Do it again for 50ps from scratch?

This practice is catching on; if you have the time.. My experience with
changing the size of monovalent cations in DNA is that owing to the 
simplicity of the perturbation, getting, say, <5% hysteresis is sufficient. 
But for big charge perturbations, this may not apply.

Another note: don't use your Gibbs parm topology file with the md program. 
Another paper: Straatsma & Berendsen, J. Chem. Phys. 89:9, 5876-5886,
1988 (ion charge perturbations in water).

Bill Ross

From chemistry-request@ccl.net Wed Jun 19 13:52:36 1991
Date: Wed, 19 Jun 1991 12:34 EDT
From: "DOUGLAS A. SMITH" <FAX0236@uoft02.utoledo.edu>
Subject: AM1 Sulfur
To: chemistry@ccl.net
Status: R

A short while back Andy Holder posted the reference for AM1 Sulfur parameters.
Would someone please repost that reference, in full if possible (I need it
for a paper).  Thanks.

Doug Smith
University of Toledo

From chemistry-request@ccl.net Wed Jun 19 14:11:32 1991
Date: Wed, 19 Jun 91 09:41 CST
From: "Mike Mitchell, UofH Chemistry, (713)-749-4229"
Subject: Re:  Free-energy perturbation calcs of zinc<->peptide binding
To: CHEMISTRY@ccl.net
Status: R




   Alan-

  I pretty much concur with what Bill Ross had to say about your problem.
You might want to check out the current JACS for the Kollman paper on
Carbonic Anhydrase wherein they paramaterize Zn:

  Merz, Murcko and Kollman in J. Am. Chem. Soc. 1991, 113, 4484-4490.

  The Straatsma paper of 1988 shows that a system consisting of an ion
in discrete waters, wherein a substantial charge is mutated (in/out), *can*
achieve convergence in ~40 ps.  This is by no means the case when VDW terms
alone or in conjunction with slight charge changes are mutated.  See:

  Mazor and Pettitt in Molecular Simulation 1991, 6, 1-4
and
  Mitchell and McCammon in J. Comp. Chem. 12, 2, 271-275.

   You may need to do lots more simulation than what you had planned.

   I'm not sure what to make of the JACS paper mentioned above.  I find
that the "slow growth" method is far less efficient for free energy calcs.
than the perturbation-windowing method, pretty much at odds with what is
stated in the paper.  I think that the jury is still out on this point.


   Mike Mitchell




From jkl@ccl.net Wed Jun 19 14:16:05 1991
Date: Wed, 19 Jun 91 14:19:26 -0400
From: jkl@ccl.net
To: chemistry@ccl.net
Subject: Re: LabVision $2500
Status: R

I am (jkl@ccl.net) forwarding commercial announcement since it is short:
Tripos' Computer-Aided Molecular Design software program, LabVision, is only
$2500 until September 30, 1991.  This is an introductory price and after Sept.
30 LabVision will cost over $10,000. LabVision is designed specifically for use
by bench scientists and educators.  Its intuitive interface allows even the
casual user to easily build, visualize, and compare molecules. The main benefit
to scientists is that they can streamline their synthesis (e.g. by computing
spatial relationships between functional groups) to make compounds which have a
higher probability of having target properties.  Scientists can readily compute
structures which have low energies and uncover conformations which are more
likely to exist.  Interfaces with programs such as MOPAC and the Brookhaven
Protein Data Bank are provided.  High quality presentation capabilities include
support of both color and black and white PostScript output.  Available on
Vax/VMS, SGI, ESV, and IBM all with Mac/PC extensions.
Please contact Kathy Clark at 800-323-2960 or by E-mail (kathy@tripos.com).


From chemistry-request@ccl.net Wed Jun 19 14:24:02 1991
Date: Wed, 19 Jun 91 13:35:09 -0400
From: zheng@retina.chem.psu.EDU (Ya-Jun Zheng)
To: jkl@MPS.OHIO-STATE.EDU
Subject: AM1 Sulphur parameters
Status: R

 The AM1 sulfur parameters are in paper by Dewar and Yuan.
  Dewar, M. J. S.; Yuan, Y. C. Inorg. Chem. 1990, 29, 3881.
 Sulfur is now available in both MOPAC 6.0 and AMPAC2.1 programs.

          Ya-Jun Zheng
          Penn State University


From jkl@ccl.net Wed Jun 19 15:52:01 1991
Date: Wed, 19 Jun 91 15:55:25 -0400
From: jkl@ccl.net
To: chemistry@ccl.net
Subject: Shake algorithm in C, anyone ?
Status: R

I (jkl@ccl.net) am forwarding this message to the list
---------- Begin Forwarded Message ----------
> Date: Wed, 19 Jun 91 12:39:05 -0700
> From: raman@thales.scripps.edu (K. Ramnarayan)
> To: chemistry-request@ccl.net

Hello,

Does anybody have information as to whether
a C version of SHAKE algorithm(used in MD)?
I am planning to convert a fortran routine, and
I can skip that if there is already one available.
Thanks much for your help.

raman
raman@scripps.edu
----------- End Forwarded Message -----------


From chemistry-request@ccl.net Wed Jun 19 16:32:43 1991
Date: Wed, 19 Jun 91 14:34:44 CST
From: dmschn@crygtw.monsanto.com
Subject: Raster-3D
To: chemistry@ccl.net
Status: R

I believe a message was posted recently about Raster-3D and anonymous ftp.
What version is this? Could someone send a copy of this message to
 mvgrie@mtcamm.monsanto.com
Thanks,
              Dora Schnur
              dmschn@monsanto.com

From chemistry-request@ccl.net Wed Jun 19 17:14:04 1991
Date: Wed, 19 Jun 91 15:01:12 EDT
From: "David A. Pearlman" <dap@predator.vpharm.com>
To: CHEMISTRY@ccl.net
Subject: Free energy perturbation calculations of zinc<->peptide binding
Status: R

>  I pretty much concur with what Bill Ross had to say about your problem.
>You might want to check out the current JACS for the Kollman paper on
>Carbonic Anhydrase wherein they paramaterize Zn:
>
>  Merz, Murcko and Kollman in J. Am. Chem. Soc. 1991, 113, 4484-4490.
>
>  The Straatsma paper of 1988 shows that a system consisting of an ion
>in discrete waters, wherein a substantial charge is mutated (in/out), *can*
>achieve convergence in ~40 ps.  This is by no means the case when VDW terms
>alone or in conjunction with slight charge changes are mutated.  See:

Following up further on the length of simulation required to reach convergence,
I direct you to two papers by Peter Kollman and myself.

1) D.A. Pearlman & P.A. Kollman (1989) J. Chem. Phys. 90, 2460-2470.
2) D.A. Pearlman & P.A. Kollman (1991) J. Chem. Phys. 94, 4532-4545.

In the first, we compare a new technique for performing FEP, "dynamically
modified windows", with simulations run using slow growth. We show the latter
does a comparitively poor job of calculating solvation free energies for
noble gases in for short time scales.

In the second, we perform a systemaic investigation of simulation time
vs. calculated solvation free energy for methane, and show that at least
200ps simulations are required to reach convergence for this very simple
system.


>   You may need to do lots more simulation than what you had planned.

Yes. See, particularly, reference 2.

>
>   I'm not sure what to make of the JACS paper mentioned above.  I find
>that the "slow growth" method is far less efficient for free energy calcs.
>than the perturbation-windowing method, pretty much at odds with what is
>stated in the paper.  I think that the jury is still out on this point.

The jury is definitely still out. Slow growth gives lower apparent
intra-run hyestesis values, but as we have shown (e.g. ref. 2 above) these
values are basically worthless as an estimate of error for slow growth
runs. In other work, we have shown that the implicit assumption of slow
growth--that the system remains in approximate equilibrium at each step--is
probably erroneous (Pearlman & Kollman (1989) J.Chem. Phys. 91, 7831-7839).
My gut feeling is that either dyanmically modified windows or theormodynamic
integration is probably the "best" method in most cases. But there's no
definitive data to support this conclusion.

>
>
>   Mike Mitchell
>

dap
===============================================================================
    David A. Pearlman

Vertex Pharmaceuticals Inc.                     email: dap@vpharm.com
40 Allston St.
Cambridge, MA 02139-4211                "Not just a job, an adventure..."
     (617) 576-3111
FAX: (617) 576-2109                              Drugs R Us...
===============================================================================