Re: [AMBER] residue-wise SASA calculation in CPPTRAJ

From: Jason Swails <jason.swails.gmail.com>
Date: Tue, 26 Mar 2013 07:54:31 -0400

On Tue, Mar 26, 2013 at 6:08 AM, aneesh cna <aneeshcna.gmail.com> wrote:

> Dear Amber users,
>
> I am trying to do residue-wise SASA calculation using 'surf' command in
> cpptraj. Below is the script I used for the calculation.
>
> *******************************************************************
> trajin md3.crd
>
> strip :WAT,:Na+
>
> center :1-293 mass origin
> image origin center
>
> surf res25 :25&!.CA,C,O,N,H out SASA.dat
> surf res64 :64&!.CA,C,O,N,H out SASA.dat
> surf res65 :65&!.CA,C,O,N,H out SASA.dat
> surf res68 :68&!.CA,C,O,N,H out SASA.dat
> surf res69 :69&!.CA,C,O,N,H out SASA.dat
> ******************************************************************
>
> But, for few residues, calculated SASA have negative values. Later, I tried
> doing residue-wise SASA for a single snapshot, using 'molsurf', 'surf'
> methods implemented in cpptraj as well as using NACCESS (
> http://www.bioinf.manchester.ac.uk/naccess/) also. For residue 68, 'surf'
> gives -9.497, 'molsurf' gives 139.179 and NACCESS gave 0.22. Considering
> all these facts, I have few queries about the residue-wise SASA
> calculation.
>
> 1. Why 'surf' calculation showing negative value for SASA? what could be
> the meaning of a negative SASA values?
>

In my opinion, it's an inadequacy of using the LCPO method to calculate
these kinds of surface areas (i.e., decomposed surface areas). Obviously
the SAs should never be negative, and the fact that they are is simply an
artifact of the P1, P2, P3, and P4 fitting parameters of LCPO. The method
was parametrized to calculate total surface areas, so it doesn't
necessarily surprise me that it doesn't work well when decomposed like this.

2. Why there is a huge difference between 'surf' and 'molsurf' SASA
> outputs? (-9.497 vs 139.179)
>

Because I believe that the molsurf method does _not_ calculate a decomposed
surface area. That is, the surface area of a mask is calculated as though
no other atom exists. You can verify this by comparing the output of:

molsurf prestrip :25&!.CA,C,O,N,H out surf.dat
strip !(:25&!.CA,C,O,N,H)
molsurf poststrip :* out surf.dat

and compare those two columns.

3. Which one of these values are trustable?
>

Definitely not LCPO and molsurf (since molsurf is not doing what you think
it is doing). Naccess appears to be designed specifically to do this --
does the value seem reasonable?


> 4. Is there a better way (or any other programs) to calculating
> residue-wise SASA over the trajectories?
>

Naccess appears to be a reasonable route. I have also used VMD in the past
using this script:
http://www.ks.uiuc.edu/Research/vmd/mailing_list/vmd-l/att-8424/_sasa_res.tcl

It is pretty slow (because it calculates the total surface area for each
residue for each frame, printing out only the contribution from each
residue). If you only want a couple residues, you can modify the script to
do that. In any case, you can use it on a couple frames to compare with
the results from Naccess.

HTH,
Jason

-- 
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Mar 26 2013 - 05:00:03 PDT
Custom Search