Database Systems
Protein Database and Molecular Surfaces
For the docking application, a huge amount of data from various sources in diffe
rent formats has to be supplied. The fundamental data are the structural data as available from the Brookhaven Protein Data Bank (PDB) in an ASCII text file format. In order to manage these 3D protein data, we designed an object-oriented
protein database system [Sommer94]. Since the complementarity of surfaces plays an important role for the protein docking application, an appropriate representation for molecular surfaces is required [KSS95]. Our surface model is based on the Connolly molecular surface (cf. figure 1) which consists of convex, saddle-
shaped, and concave patches.
Figure 1: Surface structure of the heme group of hemoglobin [Connolly83]
When implementing a system for 1:n-docking prediction, the use of database tech
niques causes a twofold benefit. First, the integrated storage of all required data in a single database system has great advantages compared to the storage by a simple file system. Second, in order to store molecular surfaces, we developed the Triedge data structure [SK95] that provides two advantages. First, a set of operations supports navigation on the molecular surface. Starting from any given position p on the surface, a neighborhood query yields the surrounding surface segment which fulfills a user-defined similarity or distance condition. Second, storage space is saved by a reduction of redundancy which is achieved by exploiting the strong regularity of the molecular surface structure.
HOME Pages:
DBS
Institute
LMU
Bei Problemen oder Vorschlägen wenden Sie sich bitte an:
wwwmaster@dbs.informatik.uni-muenchen.de
Last Modified: