From: Sébastien Légaré (Sebastien.Legare_at_rsvs.ulaval.ca)
Date: Wed Apr 15 2009 - 22:06:14 CDT
Salut Sebastian
> > I would also like to talk about one particular problem with carboxylic
> > acids in FEP. Because of the high energy barrier between syn and anti
> > conformations of the carboxyl hydrogen, the protonated state gets stuck
> > in the unfavorable, anti conformation, if the dummy hydrogen went to anti
> > while unprotonated.
>
> so for GLUP, the syn/anti conformations are defined through the dihedral
> HE2-OE2-CD-CG, (syn=0-90degs,anti=90-180degs), right?
>
Yes, the dihedral defined by HE2-OE2-CD-CG is the good one. But, it seems that
the syn/anti conformers are usually defined from the HE2-OE2-CD-OE1 dihedral.
Hence, the favorable conformation (syn or cis or Z) is at 180 degrees for
HE2-OE2-CD-CG and the unfavorable conformation (anti or trans or E) is 0
degrees.
> If yes, I see my protonated GLUP side chain in my FEP-residue E2EP
> (=E2B) in the unfavorable anti conformation (=180degs.), right after
> running VMD psfgen.
>
If your simulation starts with HE2-OE2-CD-CG at 180, that is the favorable
conformation. Everything is fine unless it goes to 0 degrees during you FEP
and does not come back rapidly.
> Your concern is also important for doing simple GLUP patches in a plain
> MD run, right? I did some tests, and found that the GLUP patch leaves
> the HE2-OE2-CD-CG dihedral in the unfavorable anti-conformation. I bet
> even long equilibration runs will not fix this issue!
>
I did not think about the effect of this for plain MD runs. I would guess that
it is important only if you really care about properties of this particular
GLUP, like we do now. But I do not think it will have a strong impact on the
rest of the system. I might be wrong.
I made some FEP simulations on ASPP residues too. Looking back at it, I see
that this problem is more present with ASPP. I do not know why. But the
important part is that if you only care about GLUP and you want to do
10-20 ns FEP simulations, there are good chances you will never
encounter this problem.
> I this concern you mentioned noted somewhere in the topology
> documentations?
>
Not that I know. I have not seen it in papers about pKa calculations from FEP,
maybe because they are often performed on systems with large pKa shifts (like
10 pKa units) and that the error coming from getting stuck in anti
conformation becomes negligible. Papers talking about this usually are about
constant pH simulations and propose various workarounds. One strange thing is
they do not agree on the energy of the syn-anti transition barrier given by
the charmm22 parameters. I really thought the energy barrier was 8.2 kcal/mol
since HE2-OE2-CD-OE1 is also assigned with the dihedral parameter
X CD OH1 X 2.0500 2 180.00
But some say it is 2.05 kcal/mol (Proteins. 56, 738-752), others say 4.1
kcal/mol (Biophys. J. 89, 141-157). I am confused.
You can look at the workarounds proposed in those two last refs or in J.
Comput. Chem. 25, 2038-2048. Personnally, I start in syn and change the
dihedral parameter to
X CD OH1 X 3.0000 2 180.00
The method including tautomerism could be nice too, but I did not look at it
enough to know if it can easily be applied to FEP.
Regards
Sébastien
> Thanks again,
> Sebastian
>
> > This have a strong effect on the computed free energy. Some workarounds
> > were proposed. The one I prefer is to raise the barrier to 6.0 kcal/mol
> > and start in syn conformation to ensure that this conformation is
> > conserved even in the unprotonated state as proposed by J. Khandogin
> > (Biophys. J. 2005, 89, 141-157), see "Protonation state models"
> > section.
> >
> >
> >
> >
> > Is it safe to say that raising this barrier does not affect the computed
> > free energy otherwise than by ensuring a syn conformation?
> >
> > Regards
> >
> > Sébastien
> >
> > On April 7, 2009 08:15:19 pm Sebastian Stolzenberg wrote:
> >> Dear All,
> >>
> >> I am trying to write my own FEP mutation: GLU -> GLUP.
> >>
> >> Attached you will find my first attempt (the "B" in "E2B" I defined for
> >> "GLUP"). Is the logic for this right? (I don't expect you guys to look
> >> for typos).
> >>
> >> Is there a nice script doing these things in a breeze?
> >>
> >> Cheers,
> >> Sebastian
This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:52:37 CST