Information extraction from 2D electrophoresis images

Two-dimensional polyacrylamide gel electrophoresis (2DE) is an important and fast growing method for the separation, identification and quantification of proteins in tissue samples. 2DE separates proteins in a sample simultaneously by molecular charge and the number of amino acids, resolving hundreds or thousands of separate protein “spots” in a single image. Discovering information (differences) between images is extremely difficult because of the large number of spots, variations in spot location, shape, and size, and differing backgrounds caused by differences in gel formation, temperature, time-voltage trajectories, laboratory equipment, and other process uncertainties. The problem is compounded when looking for trends across a large number of images. We have developed a computer algorithm that is able to automatically calibrate 2DE images relative to a standard image, thus removing the effect of variations in gel processing and image formation. An output of the process is the identification of protein spots in the image and the correspondence between the spots in different images. The algorithm can then identify consistent differences across sets of images. That algorithm is described here and is demonstrated on a series of 2DE images taken from rats under laboratory conditions