vote up 1 vote down
star

Hi,

given a protein structur from the PDB, I would like to generate NS spheres of radius Rs which cover totally the protein surface. Given RS there, NS is the maximum number of spheres so they do not overlap. I would need the coordinates of the center of each sphere.

Does anybody know if this has been implemented in some method / program? Or how to do it with scripting.

Thanks

flag
Asdf, can you add a DOI of the paper describgin the methods you are talking about? – Egon Willighagen Jan 7 at 18:43
Upvoted, to get asdf past the 21 minimum, so that he can edit his question... – Egon Willighagen Jan 7 at 18:44

closed as no longer relevant by Egon Willighagen♦♦ Feb 18 at 16:30

2 Answers

vote up 0 vote down

Your question is vague and I don't understand the underlying reason for doing this. Some PDB structures require additional changes to make them be in biological form, like dimers which are stored in the PDB only as a monomer. By limiting the structures to protein do you mean to take away anything which isn't a protein, including waters, DNA, ligand, and covalently attached structures? What if the structure contains multiple proteins? Multiple chains covalently bound with an S-S bond? How do you want to handle missing structures?

Which surface do you mean? The molecular surface, the solvent accessible surface, or something else? How do you want to handle pores and cavities? Consider in the extreme case the full shell of a hollow virus capsid, containing no DNA or RNA.

Does "cover" mean each sphere is completely outside of the surface or can it intersect the surface? Or that the center of the sphere must not be inside the surface of the protein? Can the spheres intersect each other?

It's impossible to specify both N and R and get an answer which covers an arbitrary surface. If N = 1 then R must be at least 1/2 the maximum width of the protein, if the sphere can be placed in the center. Otherwise there is no solution.

My guess is you're trying to do some sort of structural domain identification, but in that case having spheres is too limiting. I've seen commercial work (related to electrostatic surfaces) which use Gaussian ellipsoids. However, it's been too long since I've been in that field to know which tools are available.

link|flag
vote up 0 vote down

Isn't this the Van der Waals surface is defined? It uses the outside surface of the spheres. See code in Jmol, I guess.

link|flag
The vdW surface doesn't have the concept of a specific count of non-overlapping spheres. – Andrew Dalke Jan 23 at 1:54
True - I didn't read the question right. – baoilleach Jan 23 at 12:23
Yeah, well neither did I in my response. The question really needs clarification since I can't figure out what the OP wants. – Andrew Dalke Jan 24 at 3:06

Not the answer you're looking for? Browse other questions tagged or ask your own question.