In the project BIOWEPRO (1993 - 1996, funded by the German Ministry for Education, Science and Technology) we use techniques of pattern analysis to solve the docking problem. Within a hierarchical system we compute potential docking candidates and docking positions before subsequently the exact free energy of the complex is computed. The main idea is to detect geometrical and chemical complementarities between surfaces of involved proteins and from these to estimate potential docking positions. A prerequisite for this is the development of a suitable computer representation of proteins and their surfaces. Such a representation promotes the segmentation of protein surfaces in characteristic regions. These in turn are the basis to determine potential docking positions using a knowledge based approach with the semantic network ERNEST. These initial estimates are further refined employing a vector valued cross correlation of the sampled proteins. Finally we develop an initial approach for the selection of docking candidates for 1:n-docking using a polynomial classifier.
Partners within Biowepro
|sneumann@TechFak.Uni-Bielefeld.DE||Last Update: 18 Dec 02|