From: L. Michel Espinoza-Fonseca (mef_at_ddt.biochem.umn.edu)
Date: Sat Nov 10 2007 - 19:29:31 CST
To do science one should be (according to my very personal point of
view) as open as possible to accept the fact that what one does is
basically our best available approximation. Even "in vitro" and "in
vivo" experiments are approximations, and we have been relying on that
approximate knowledge to solve a problem. This is not "fudging". If
you become very orthodox, then (I say it in a very respectful way, so
please don't get offended) you should find something else that you
might not consider as "fudging". I repeat, science is just an
approximation and we try to do our best to get it improved.
Meeting experimental results is part of our job as theoretical
biochemists/biophysicists. There is a nice paper by Ruth Nussinov
where she nicely describes how important is to make experiments and
theoretical calculations very good (and necessary) partners. And this
is not about pleasing reviewers; it is more about showing to other
scientists that we are able to integrate the bunch of data with the
simulations and come up with a solution to improve what has been done.
By doing that, at the end you'll be the one who is the most pleased.
Finally, and in agreement with Richard Law, if you're unable to
"please" the reviewers, the you won't be able to publish any work, get
grants and continue doing your research. That's a very important part
of science: Learning from others (the reviewers, for instance) and
sharing your knowledge by publishing.
Michel
2007/11/10, Richard Wood <rwoodphd_at_yahoo.com>:
>
> Your model IS constructed based on experimental data, and during your
> calculation, it changes. So, you fix it to meet "experimental results",
> which is what the reviewers want. That's fudging, in my book.
>
> I've also mentioned previously my concern about changing ensembles after
> equilibration. Perhaps that is why the volume changes on going from NVT to
> NPT, again to please reviewers. I was just at a CHARMM tutorial where we
> were told to "never simulate using NVT because experiments are always at 1
> atm". I think changing horses in mid-stream leads to problems, but that is
> my opinion, I guess. I would also guess that the constant P piston is what
> is causing the membrane to compress.
>
> Richard
> Richard L. Wood, Ph. D.
> University of Minnesota
> Dept. of Medicinal Chemistry,
> College of Pharmacy
> 717 Delaware St. SE
> Minneapolis, MN 55414-2959
> woodx278_at_umn.edu
>
>
> ----- Original Message ----
> From: Himanshu Khandelia <hkhandel_at_memphys.sdu.dk>
> To: Richard Wood <rwoodphd_at_yahoo.com>
> Cc: namd-l_at_ks.uiuc.edu
> Sent: Saturday, November 10, 2007 3:20:47 AM
> Subject: Re: namd-l:
> NPT_NPgammaT_or_NPAT_for_membrane_with_CHARMM_forcefield
>
> No, this is not fudging results. This is modeling and simulation, and any
> model is constructed based on experimental data.
>
> > > Isn't this just "fudging" results? It appears to me that you're just
> > > changing your results so they fit experimental ones by using one method
> to
> > > equilibrate, and then another to do your production run. I thought
> science
> > > was about learning new things, and not "pleasing" reviewers.
>
> Its not about pleasing reviewers. Its about doing it the best method
> currently available. (Un)fortunately, often in science, there is no
> consensus about one single best method.
>
> > I would think one would want to do equilibration using the same method
> > that one does for the production run, as I would think this would be one
> > thing that reviewers would be more concerned about than one's area per
>
> Why should one equilibrate and run dynamics using the same ensemble ?
> Equilibration is only to obtain a local energy minimum, so that the
> simulation is stable during equilibrium sampling. As long as one is doing
> equilibrium sampling in a single thermodynamic ensemble in the phase
> space, who cares how the equilibration was done ?
>
>
> > I would also think that if the area per lipid shrinks, one could
> > start with a larger lipid so that when it shrinks during equilibration, it
> > would be close to the "correct size".
>
> Good idea, but the area per lipid does not shrink during equilibration, it
> keeps shrinking during equilibrium dynamics. A variation of what you have
> suggested has been mentioned in the list before. Check this out:
> http://www.ks.uiuc.edu/Research/namd/mailing_list/namd-l/6861.html
>
> -Himanshu
>
>
>
> __________________________________________________
> Do You Yahoo!?
> Tired of spam? Yahoo! Mail has the best spam protection around
> http://mail.yahoo.com
This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:45:31 CST