# Re: [AMBER] PCA Calculations

From: Daniel Roe <daniel.r.roe.gmail.com>
Date: Tue, 30 Jun 2015 09:44:44 -0600

Hi,

On Mon, Jun 29, 2015 at 8:43 PM, Parker de Waal <Parker.deWaal.vai.org>
wrote:

> In this PCA, I am trying to identify stable interactions with a
> phosphoserine :335, found on protein A, and protein B :350-650.
>

It's not clear to me how PCA is going to help you identify stable
interactions in this case. When I hear stability, it makes me think of free
energy - a "stable" interaction is one that has a high population in a
well-converged ensemble. PCA in my mind is more of a tool for analyzing
dynamics, i.e. the kinds of motion that are present in your system.
Performing PCA on a single residue will give you information on how that
residue moves within the specified reference frame, but I'm not clear on
how it shows you "stability" (maybe I'm missing something).

> Specifically, I was wondering how one selects the number of vecs for both
> the analyze matrix and projection commands.
>

You need to look at the eigenvalues you get from diagonalization of your
coordinate covariance matrix; this gives you an idea of how much each
eigenvector contributes to the overall motion. The 'eigenval' keyword of
the 'modes' analysis command in cpptraj will create an "eigenvalue
fraction" plot which will give you an idea of how much each eigenvector
contributes. In some systems the first two PCs are responsible for a large
majority of the motion, but this is by no means always the case. In some
cases the first 3 or 4 modes may even be degenerate - you just won't know
unless you look at your eigenvalues.

Hope this helps,

-Dan

> Using this script above, I am generating 25 eigenvectors, but only
> projecting on the first 3. As someone that is new to MD PCA, does this
> sound appropriate?
>
> Finally, from the projection output, how does one decide which modes to
> plot and obtain kcal/mol as seen here:
>
> Any feedback would be greatly appreciated!
>
> Best,
> Parker
> _______________________________________________
> 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 Tue Jun 30 2015 - 09:00:03 PDT
Custom Search