Two-dimensional electrophoresis image interpretation.

Abstract

A novel method to interpret two-dimensional electrophoresis gels is presented. Genetic background and electrophoretic processes are summarized. Present methods to analyze gel images and to exploit series of gels are described, then their drawbacks are outlined. Artificial Intelligence techniques are introduced to build an image interpretation system which can compensate for certain failures of present methods and augment their efficiency. By reproducing methods of biological experts, this system automatically identifies proteins--whether isolated or inside constellations--on an electrophoretic gel. This system is based on a modular architecture, featuring image processing procedures, which allows extraction of parameters on the image and a top-down and bottom-up reasoning process. The reasoning process first matches extracted parameters to possible geometric models of proteins; it then returns to the image to determine possible missing elements on the gel. A prototype of this system was implemented and tested on plasma gels to identify apolipoproteins.