Vohradský J, Pánek J
Czech Academy of Sciences, Institute of Microbiology, Praha.
Electrophoresis. 1993 Jul;14(7):601-12. doi: 10.1002/elps.1150140195.
A computer-aided quantitative method for a complex analysis of gel electrophoretograms is presented. The analysis consists of several steps: (i) determination of the background image by methods of mathematical morphology and its subtraction from the gel image, (ii) selection of an appropriate part of the gel lane including curved lanes and lanes with a nonuniform width, (iii) computation of the lane densitogram by averaging several lane-parallel scans, (iv) decomposition of the lane densitogram into component bands using a data selecting algorithm and Marquardt's minimizer. Several different functions for component bands are utilized. It is shown that the densitogram can be decomposed into component bands with reasonable accuracy only if an appropriate model function is chosen. The algorithms are tested on several different gel electrophoretograms which show typical features as a nonuniform background, curved lanes, an asymmetrical band shape and a superposition of small bands on the shoulders of big ones. It is shown that overlapped bands are best approximated by an asymmetrical Gausian curve and an asymmetrical Gauss-Cauchy function. Linear response to the serial dilution of the protein sample is tested.
提出了一种用于凝胶电泳图谱复杂分析的计算机辅助定量方法。该分析包括几个步骤：(i) 通过数学形态学方法确定背景图像并从凝胶图像中减去；(ii) 选择凝胶泳道的适当部分，包括弯曲泳道和宽度不均匀的泳道；(iii) 通过对几个与泳道平行的扫描进行平均来计算泳道光密度图；(iv) 使用数据选择算法和Marquardt最小化器将泳道光密度图分解为组分条带。使用了几种不同的组分条带函数。结果表明，只有选择合适的模型函数，光密度图才能以合理的精度分解为组分条带。这些算法在几种不同的凝胶电泳图谱上进行了测试，这些图谱显示出典型特征，如不均匀背景、弯曲泳道、不对称条带形状以及小条带在大条带肩部的叠加。结果表明，重叠条带最好用不对称高斯曲线和不对称高斯-柯西函数近似。测试了对蛋白质样品系列稀释的线性响应。
