VMD-L Mailing List

From: John Stone (johns_at_ks.uiuc.edu)
Date: Fri Jun 18 2010 - 15:25:51 CDT

Nuno,
  Yes, I would expect it to be an easy thing to write in CUDA.
This isn't on my TODO list currently, as I'm finishing several
other GPU kernels currently. If you wanted to work on it, I would
be happy to give you tips/guidance as you make progress.

Cheers,
  John Stone
  vmd_at_ks.uiuc.edu

On Fri, Jun 18, 2010 at 09:20:23PM +0100, Nuno Sousa Cerqueira wrote:
> Hello,
> I have just come from a summer school on GPU computing.
> I was wondering if the SASA algorithm would be a nice code to implement in
> CUDA?
> Is there any work going in this direction?
> Regards,
> Nuno
> Begin forwarded message:
>
> From: John Stone <johns_at_ks.uiuc.edu>
> Date: June 18, 2010 8:22:19 PM GMT+01:00
> To: Edward Lyman <edward.lyman_at_gmail.com>
> Cc: vmd-l_at_ks.uiuc.edu
> Subject: Re: vmd-l: -samples flag of measure sasa
> Edward,
> Yes, the 'atom' sphere sizes are the atom's radius plus the probe
> radius,
> as you can see on line 1132 of the code there. Yes, 500 samples are
> used
> by default. The code doesn't treat any atom type specially, it applies
> the same rules to all. All of the radii are specific to each atom, so
> it just uses what it's given. You can manipulate how it behaves by
> changing the radii of your atoms, or by selecting or not selecting some
> atoms.
>
> Cheers,
> John Stone
> vmd_at_ks.uiuc.edu
>
> On Fri, Jun 18, 2010 at 01:15:10PM -0600, Edward Lyman wrote:
>
> Thanks very much, that is useful information: area/point = (surface
> area
>
> of the atom)/(number of randomly sampled points).
>
> the surface are of the 'atom' is actually the area of the sphere
> with a
>
> radius = (atom radius + probe radius), right?
>
> It looks from the code that 500 random points are used...is that
> correct?
>
> So...*IF* I could assume that every atom considered in the SASA
>
> calculation has the same radius, I would know the area/point, and I
> could
>
> easily calculate the area for each separate patch.
>
> one last question: are the H's and the C's considered seperately for
> the
>
> purpose of the SASA calc? or do you just puff out the radius of the
>
> carbons a little bit and save some calculation time?
>
> Thanks very much for the help.
>
> Ed
>
> On Fri, Jun 18, 2010 at 12:10 PM, John Stone <johns_at_ks.uiuc.edu>
> wrote:
>
> Hi,
>
> The -samples flag controls how many random sample points are used
>
> to estimate the surface area for each atom. The points are
> distributed
>
> randomly over the sphere, and the area is estimated by multiplying
> the
>
> fraction of samples that are exposed by the surface area of the
> sphere
>
> for the atom under consideration. When you retrieve the actual
> list of
>
> sample points, you are getting the points that passed the solvent
>
> accessibility test for the combination of selection and/or
> restricted
>
> selection you provided. Does that answer your question?
>
> You can see the source code that implements the calculation in the
>
> measure_sasa() routine here:
>
> http://www.ks.uiuc.edu/Research/vmd/doxygen/Measure_8C-source.html#l01058
>
> (that URL will change with time, but it's good for the short-term)
>
> Cheers,
>
> John Stone
>
> vmd_at_ks.uiuc.edu
>
> On Thu, Jun 17, 2010 at 10:47:34AM -0600, Edward Lyman wrote:
>
> Hi all,
>
> I am looking for an explanation of the -samples flag for measure
>
> sasa
>
> (somehow it has been omitted from the manual). I guess it is
> simply
>
> the
>
> density of points that is used to sample the surface, but I would
>
> like to
>
> know what is the default value, and what the value means. For
>
> instance,
>
> does each point represent a (roughly) constant area? Or is each
>
> point a
>
> vertex in a more complicated (eg, Delaunay) triangulation, and so
>
> could
>
> represent a wide range of areas?
>
> My interest stems from the following problem: I have used "measure
>
> sasa"
>
> to identify several distinct patches of solvent accessible
> residues
>
> on the
>
> surface of a protein. I can store those points, and then post
>
> process them
>
> using a simple (breadth-first) search algorithm to automate the
>
> process of
>
> distinguishing different patches. Now I want to know the area of
>
> each
>
> patch. If each vertex represents a constant area...problem solved.
>
> But if
>
> the surface is more complicated...the problem becomes considerably
>
> trickier, and I then have to recover the (nonoverlapping)
> triangles
>
> in
>
> each patch and add up their areas.
>
> Thx,
>
> Ed
>
> --
>
> NIH Resource for Macromolecular Modeling and Bioinformatics
>
> Beckman Institute for Advanced Science and Technology
>
> University of Illinois, 405 N. Mathews Ave, Urbana, IL 61801
>
> Email: johns_at_ks.uiuc.edu Phone: 217-244-3349
>
> WWW: http://www.ks.uiuc.edu/~johns/ Fax: 217-244-6078
>
> --
> NIH Resource for Macromolecular Modeling and Bioinformatics
> Beckman Institute for Advanced Science and Technology
> University of Illinois, 405 N. Mathews Ave, Urbana, IL 61801
> Email: johns_at_ks.uiuc.edu Phone: 217-244-3349
> WWW: http://www.ks.uiuc.edu/~johns/ Fax: 217-244-6078
>
> Phd, Nuno M. F. Sousa A. Cerqueira
> Assistant Researcher
> ------------------------------------------------------------
> Chemistry Department, Faculty of Sciences
> University of Porto
> Rua do Campo Alegre, s/n
> 4169-007 Porto - Portugal
> Tel: +351 220402501
> Fax: +351 220 402 009
> ------------------------------------------------------------
> http://www.fc.up.pt/pessoas/nscerque 

-- 
NIH Resource for Macromolecular Modeling and Bioinformatics
Beckman Institute for Advanced Science and Technology
University of Illinois, 405 N. Mathews Ave, Urbana, IL 61801
Email: johns_at_ks.uiuc.edu                 Phone: 217-244-3349
  WWW: http://www.ks.uiuc.edu/~johns/      Fax: 217-244-6078