Methods for Computer Analysis and Comparison of Two-Dimensional Protein Patterns Obtained by Electrophoresis

摘要

Automated analysis of gel images is an important part of the quantitative study of proteins that have been separated by two-dimensional gel electrophoresis. In order to establish the sensitivity of assays relying on this separation technique it is crucial to know how accurately and reliably information can be extracted from gel images. We have examined some of the main factors that influence the results of quantitative gel image analysis. These include the choice of parameters for image processing procedures to remove high and low spatial frequencies, and the selection of appropriate methods for analysis as well as comparison. From our experiments with simulated images as well as gel images we conclude that an analysis procedure that does not rely on assumptions regarding the greyvalue distribution within spots, i.e., Gaussian spot profiles, appears to result in slightly more accurate and significantly more reproducible spot volume estimates than a parametric method that involves suboptimal surface fitting. We have also seen that analysis results can be quite sensitive to image processing parameters. CVs of up to 50% were observed when the window size for our background subtraction filter was varied within a range selected according to the rule of thumb to choose the window larger than the biggest spot in the image. 20 refs., 3 figs. (ERA citation 11:041668)