Defining Exposed Residues From Dssp Solvent Accessibility Value

Question

[User 1933]

DSSP provides solvent accessibility value for each residue. Base on this value one can find buried residues easily. (this value is almost 0) but for being expose, It can be depends on the our threshold, say above 100A is exposed.

My question basically is about a nice way to choose best threshold ! Is it only the matter of taste ?!

I am dealing with 200 proteins from 7 families which their structures are derived from 7 different pdb structure. I am aiming to see which AA are most on the surface.

May be for some my question sounds like this but Its not addressing exactly my question :-)

[Michael Schubert]

You're not by any chance confusing DSSP (define secondary structure of proteins) with RSA (relative solvent accessibility) values?

[Neilfws]

I don't think they're confused; output of DSSP includes solvent-accessible area, from which RSA can be estimated given a table of amino acid surface areas.

[Michael Schubert]

Ok, my mistake then.

Answer 1

[Neilfws]

You're correct in assuming that you need to choose a threshold below which a residue is considered "buried", or inaccessible. However, you should not use the absolute accessible area; what you need to calculate is the relative solvent accessible area, or RSA.

To do this, you'll need to parse the output from DSSP, extract the value from the ACC column and divide it by the total surface area for the residue. You can find tables with surface areas of amino acids on the Web; here's one, or search Google.

Note that these areas are often estimated assuming that the residue, X, is in a tripeptide, G-X-G. This can result in estimates of RSA that are more than 100% (i.e. the area given by DSSP for highly-exposed residues can be more than the estimated value for the residue in G-X-G).

Once RSA is calculated, you need to select a threshold - and this is quite arbitrary. Some people define RSA < 20% as "completely buried". It's probably best to read a few papers on the topic and see what most people use: this one is a good starting point.

Personally, I find the output of STRIDE somewhat easier to parse than DSSP (it does the same thing); also, you'll benefit from using an existing library for parsing (e.g. Bioperl's Bio::Structure::SecStr::DSSP::Res).

[Pierre Poulain]

One can also find values of the surface area of residue X in the tripeptide A-X-A: http://www.biomedcentral.com/1471-2105/5/51 No idea if it's "better" than G-X-G...

[User 1933]

@neilfws, FYI, bio3d has a function which parse DSSP output.