Voronoia

Description

A database for the computation of atomic packing density values of protein structures
Computation Packing & Cavities DOPP Reference Values Glossary References


Packing

Generally, atomic packing analysis quantifies, how close an atom is located to all its neighbouring atoms. To weigh each neighbor atoms' contribution in a reasonable way, the atomic volumes are calculated. This can be done in various ways. Voronoia applies Voronoi Cell [2], an extension of the original Voronoi procedure, where hyperboloid interfaces are constructed between atoms, allocating the total space within a protein structure among all atoms (fig. 1 and fig. 2).

The atomic packing density is a measure that qualifies how well a certain protein is packed. To evaluate this, the Van-der-Waals Volume V(VdW) of an atom (the space inside the atom's Van-der-Waals radius) is extended by 1.4 Angstrom, a convened value for the radius of a solvent molecule. This latter volume is called Solvent Excluded Volume V(SE). Both, the atoms' Van-der-Waals Volume (fig. 1, dark colours) and the Solvent Excluded Volume (fig. 1, bright colours) are cut by the separating surfaces with atoms in very close proximity. Only, in a perfectly packed structure, the V(SE) would be occupied entirely by separating surfaces and the packing density value would be 1.0. To estimate the packing value calculated by Voronoia the user may compare it to reference values that were calculated from defined sets of proteins.

The atomic packing density is definded as: PD = V(VdW) / [V(VdW) + V(SE)]
Fig. 1: The Voronoi Cell procedure assigns portions of space to each heavy atom (VdW Volume, dark colors, Solvent Excluded Volume, light colors) according to their radii. To do this, hyperboloid interfaces are constructed between the atoms CB, DG2, OG1. The position of that interface depends on the atomic radii in a way that the larger atoms also get the larger portions of space.      Fig. 2: The resulting compartment defined by the interfaces - the Voronoi Cell - thus consists of a mixture of convex and concave surfaces. More precisely, each of its points are from the atoms i and j exactly radius(i) + d and radius(j) + d apart.


Cavities

Protein cavities are regularly found in protein structures with more than 150 amino acids. These packing defects are defined to be large enough to enclose at least one water molecule. Cavities are totally buried locations, in which a 1.4 Angstrom virtual atom could be placed. This probe must not intersect any atoms' Van-der-Waals sphere and the cavity must not extend to the protein surface. Many of these cavities are occupied by water molecules in vivo, but often they remain undetected in crystal or NMR structures.



Fig. 3: In the Glycerol-Phospate transporter the cavities are located at functionally important protein sites. Polar cavities (red) indicate the channels pore, hydrophobic cavities, regions of putative high compressability that are located at the hinge reagions of the membrane transporter [5].




Voronoia uses regular PDB structure files as input, and creates modified PDB files containing packing densities for each atom. The positions of interior cavities, their neighbor atoms and protein surface atoms are listeded separately. Both cavities and their surroundings may be written as separate structure files. Textual reports that summarizes all this information and the deviation of packing from reference data are created for each structure. Voronoia is capable to create these reports for sets of structures in one go. This also generates average packing values, that may be used as a new reference set. Along with the program, reference data is provided for a current non-redundant set of structures based on SCOP superfamilies, and for a set of 66 transmembrane domains.


Download Reference Values


Dictionary Of Packing in Proteins

DOPP is a database that stores precalculated packing files in a data bank. The calculation of packing files at higher accuracy is a time consuming process that may last up to several minutes. To allow the user to quickly visualize the packing of a particular file, the packing of all biounits from the PDB 2007 release was calculated beforehand at high accuracy. For this purpose the PDB header was extracted and stored in a relational database to provide fast searching. On top of that a live search with a powerful query mechanism was implemented with Asynchronous JavaScript And XML (AJAX) and PHP. Furthermore, all biounit files where filtered for certain properties which do not allow the display of the results on a webpage. Thus, it is for example not possible to show large files with up to several hundred MB. In that case the user will be informed about the reason why a particular file is not available in our database.
The recent release of DOPP contains biounits. This database will be updated every six month by means of an automatic script. The next data drop will take place in February. Currently new biounit files are calculated.
To get an overview about the biountis in DOPP with resolution and methods see here.