2D Gel Matching

This project is a joint project of the Institute of Computer Science of Freie Universität Berlin and Deutsches Herzzentrum (German Heart Center) Berlin. The main topic of research are 2-dimensional gel images, that are produced by high-resolution gelelectophoresis-techniques. The gelelectrophoresis has been established to be a central molecular-biological method for the analysis of the protein/DNA-compound of tissue samples. Each ``spot'' in a gel image that has been produced by gelelectrophoresis represents one protein appearing in the sample. The analysis of the pictures helps to discover molecular and genetic reasons of heart diseases.

Until now, the interpretation of gel images is mainly based on the exact (and time consuming) examination by experienced specialists. The main goal of the project is to design and implement algorithms for the effective, computer supported gel analysis. Central to our investigations are two steps of this procedure, the gel-matching (assignment of corresponding spots from different pictures) as well as the setup and the administration of a 2-D gel protein database.

The matching represents a very important and time consuming prerequisite for the quantitative and qualitative data analysis of protein images. Within the matching, geometric distortions, that appear when producing protein samples, are to be equilibrated. The corresponding algorithmic problem is a variation of 2-dimensional pattern recognition, where the main difficulty is produced by geometric distortion. The project aims to develop new algorithms for gel matching, based on already known procedures for point pattern matching from computational geometry. Because of the fundamental nature of the task of approximative point pattern matching, every single progress within the work on this problem will also have great importance for various other applications.

More information