RE: New NAMD version 2011-03-03 uses

From: Christian Jorgensen (christian.jorgensen_at_merton.ox.ac.uk)
Date: Thu Mar 17 2011 - 12:58:03 CDT

Hi all,

when using ParseFEP with a fepout file for the forward and the backward run
I keep encountering a domain error of the form

domain error: argument not in valid range
domain error: argument not in valid range
    while executing
"expr sqrt($instant_error_gauss) "
    (procedure "::ParseFEP::FEP_formula" line 361)
    invoked from within
"::ParseFEP::FEP_formula $file $file_entropy $file_lambda $window $mean_xi $fororback"
    (procedure "::ParseFEP::normal_parse_log" line 60)
    invoked from within
"::ParseFEP::normal_parse_log $::ParseFEP::fepofile forward "
    (procedure "::ParseFEP::namdparse" line 172)
    invoked from within
"::ParseFEP::namdparse"
    (in namespace inscope "::ParseFEP" script line 21)
    invoked from within
"::namespace inscope ::ParseFEP {
        ##if {$::ParseFEP::fepbofile != ""} {set ::ParseFEP::sosindex 1 }

        if { [string length $fepofile] < 1 } {
        tk_dial..."
    invoked from within
".parseFEP.runbutton invoke"
    ("uplevel" body line 1)
    invoked from within
"uplevel #0 [list $w invoke]"
    (procedure "tk::ButtonUp" line 22)
    invoked from within
"tk::ButtonUp .parseFEP.runbutton"
    (command bound to event)

Does this invalid argument arising when calculating sqrt($instant_error_gauss) occur because $instant_error_gauss is taking values <0?
I believe it is causing me issues because the ParseFEP log file stops at λ=0.5

#=================================================
# Free energy perturbation
# --------------------------------
#forward/backward λ ∆∆A ∆A δε
#
#-------------------------------------------------
forward: 0.0000 0.0000 0.0000 0.0000
forward: 0.1000 -0.9392 -0.9392 0.0506
forward: 0.2000 -0.8597 -1.7989 0.1287
forward: 0.3000 -0.5557 -2.3547 0.1357
forward: 0.4000 -0.5542 -2.9089 0.1381
forward: 0.5000 -0.4808 -3.3897 0.1421

Any input here would be really helpful
thanks
Christian

________________________________________
From: owner-namd-l_at_ks.uiuc.edu [owner-namd-l_at_ks.uiuc.edu] On Behalf Of Chris Chipot [chipot_at_ks.uiuc.edu]
Sent: Tuesday, March 15, 2011 10:37 PM
To: Jeff Wereszczynski
Cc: namd-l_at_ks.uiuc.edu
Subject: Re: namd-l: New NAMD version 2011-03-03 uses

Hi Jeff,

in its current form, ParseFEP expects you to feed it an fepout
file for the forward simulation and possibly a second fepout
file for the backward transformation, in which case, you can
request a simple-overlap sampling (SOS) or a Bennett
acceptance ratio (BAR) estimator of the free energy. I
am inclined to advocate to use the latter, which is the
maximum-likelihood estimator. It should also yield the
smallest variance (see our Good Practices paper, J. Phys.
Chem. B 2010, 114,10235–10253).

Ideally, one would like to have a single file containing the
free-energy changes for both the λ to λ + δλ and the λ to
λ - δλ transformations generated from a single FEP run,
i.e. a double-wide sampling calculation. As Chris Harrison
pointed out, this is not yet implemented in NAMD, though
we hope to have it available in the near future. The only
advantage to have an effective implementation of the
double-wide sampling algorithm is to make the concurrent
estimation of the forward and backward free-energy changes
appreciably cheaper than two individual calculations would be.

Hence, for now, please, organize your bidirectional calculations
in two runs, one from 0 to 1 and the other from 1 to 0. One
last, trivial comment. The rewriting of the potential energy
function with a soft-core term removes the linearity of the
former in λ. As a result, the result of a backward simulation
can no longer be inferred from the sole knowledge of the
forward transformation. The forward and the backward
free-energy changes must be determined separately.

Cheers,

Chris Chipot

On 3/15/11 11:49 AM, Jeff Wereszczynski wrote:

Hi Chris,

These seem like some nice tools!

Not to hijack this thread, but I noticed this ParseFEP included a
reference to "bidirectional FEP." By this do you mean actually
running the simulations at a given lambda value twice, once with the
work functions calculated to the next window and once to the previous,
or can NAMD now calculate both the forward and reverse work functions
simultaneously? The latter would be ideal.

Thanks,
Jeff

2011/3/15 Chris Chipot <chipot_at_ks.uiuc.edu><mailto:chipot_at_ks.uiuc.edu>:

Christian,

please, check http://www.ks.uiuc.edu/Research/vmd/plugins/parsefep/, which
should do exactly what you need. ParseFEP
is part of the new version of VMD.

Cheers,

Chris Chipot

On 3/15/11 9:52 AM, Christian Jorgensen wrote:

Hi all,
I have the new NAMD version from 2011-03-03 and was wondering
if someone could briefly outline if there are any major changes
to the necessary input file used (e.g. variable name changes)?

I am specifically interested in the free energy perturbation formalism.
I have done some runs with the new version using the following input

# FEP
source ./fep.tcl
alch on
alchType fep
alchFile system.fep
alchCol B
alchOutFile out.fepout_1
alchOutFreq 500

alchVdwLambdaEnd 1.0
alchElecLambdaStart 0.5
alchVdwShiftCoeff 5.0
alchDecouple yes

alchEquilSteps 37500 ; # 75ps
set nSteps 75000 ; # 150ps

runFEP 0.0 1.0 0.0625 $nSteps ; # 10 windows outward

Also, I am interested in whether there have been developments
in how namd can deal with the error analysis of the simulation.
Is there a way of outputting ΔU values during the simulation (to
obtain an idea of the probability distribtions P(ΔU) of the simulation)?
Ultimately I wish to obtain an estimate of the simulation bias.

Thank you for any input here
Christian

_______________________________________________________________________

Chris Chipot, Ph.D.
Theoretical and Computational Biophysics Group
Beckman Institute
University of Illinois at Urbana-Champaign
405 North Mathews Phone: (217) 244-5711
Urbana, Illinois 61801 Fax: (217) 244-6078

                           E-mail: chipot_at_ks.uiuc.edu<mailto:chipot_at_ks.uiuc.edu>
                                   Christophe.Chipot_at_edam.uhp-nancy.fr<mailto:Christophe.Chipot_at_edam.uhp-nancy.fr>
                           Web: http://www.ks.uiuc.edu/~chipot
                                   http://www.edam.uhp-nancy.fr

The light shines in the darkness, and the darkness has not overcome it.
                                                             John 1:5.
_______________________________________________________________________

--
_______________________________________________________________________
Chris Chipot, Ph.D.
Theoretical and Computational Biophysics Group
Beckman Institute
University of Illinois at Urbana-Champaign
405 North Mathews                                 Phone: (217) 244-5711
Urbana, Illinois 61801                            Fax:   (217) 244-6078
                            E-mail: chipot_at_ks.uiuc.edu<mailto:chipot_at_ks.uiuc.edu>
                                    Christophe.Chipot_at_edam.uhp-nancy.fr<mailto:Christophe.Chipot_at_edam.uhp-nancy.fr>
                            Web:    http://www.ks.uiuc.edu/~chipot
                                    http://www.edam.uhp-nancy.fr
The light shines in the darkness, and the darkness has not overcome it.
                                                              John 1:5.
_______________________________________________________________________

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 05:23:41 CST