Re: [AMBER] PCA 2dprojection with xtal data and energy information

From: Daniel Roe <daniel.r.roe.gmail.com>
Date: Sat, 13 Dec 2014 16:48:58 -0700

Hi,

On Sat, Dec 13, 2014 at 3:39 PM, BERGY <nucleic81.gmail.com> wrote:
> I am not sure ho the what energy is plotted .How it should be calculated?

These are free energies calculated from histogram bin populations,
i.e. the free energy of bin i, dGi is:

dGi = -RT * ln(popi)

where popi is typically the population of bin i divided by the
population of the most populated bin (making your free energy minimum
always 0). This can be done with the cpptraj command 'hist' and the
keyword 'free <temperature>'. However, be aware that your data needs
to be fairly well converged in order to get accurate estimates of free
energy this way. You'll typically need at least 2 independent
simulations starting from different initial conditions to really
assess convergence.

> For now I have done
> aligning all the trajectories to the average structure and then calculate
> the
> projections.
> -------------
> matrix covar name covmat out covmat-ca.dat .CA
> analyze matrix covmat out evecs-ca.dat vecs 2
> projection modes evecs-ca.dat out pca12-ca beg 1 end 2 .CA
> ------------
> Tthe two averages from each of the two simulation our different..
> Can I compare the two 2D projections?

Maybe. When you perform PCA for each trajectory separately there is no
guarantee that the eigenvectors from each are the same; odds are they
will differ at least by a little, and maybe by a lot. You would have
to calculate the angle between each eigenvector to see how
well-aligned they are. A better approach in my opinion is to generate
the covariance matrix from all of your trajectories, then project
along the eigenvectors for each trajectory separately. For some
examples of this approach see some recent work from our lab (and the
supplementary information therein.):

http://www.sciencedirect.com/science/article/pii/S0304416514003092
http://pubs.acs.org/doi/abs/10.1021/jp4125099
http://pubs.acs.org/doi/abs/10.1021/ct400862k

> If I have xtal.pdb , how to get the 2p for that structure and overlay on
> the 2D projection plot?

If I understand your question correctly, you need to project the PDB
coordinates along your calculated eigenvectors. The trick is ensuring
the atoms in the crystal PDB have the same ordering as the topology
you used to calculate your covariance matrix. Your best bet for that
is to just use the initial structure and topology you got from tleap,
prior to minimization.

Hope this helps,

-Dan

>
> Thanks in advance.
>
> Best,
> Bergy
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber



-- 
-------------------------
Daniel R. Roe, PhD
Department of Medicinal Chemistry
University of Utah
30 South 2000 East, Room 307
Salt Lake City, UT 84112-5820
http://home.chpc.utah.edu/~cheatham/
(801) 587-9652
(801) 585-6208 (Fax)
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Dec 13 2014 - 16:00:02 PST
Custom Search