基于小波分析的膜蛋白跨膜区段序列分析和预测

膜蛋白是一类结构独特的蛋白质，在各种细胞中普遍存在，发挥着重要的生理功能。目前仅有少数膜蛋白听结构被实验测出，因此用计算机预测膜蛋白的结构是蛋白质结构预测的主要研究内容之一。膜蛋白一般在膜上形成保守的跨膜螺旋结构，序列特征明显，比较适合用预测的方法确定跨膜螺旋区段的位置。国际上已有一些研究者用人工神经网络方法、多序列比对方法和统计方法进行了预测尝试，取得了一定的成功经验。我们对蛋白质序列数据库中的

ANALYSIS AND PREDICTION OF PROTEIN TRANSMEMBRANE HELICALREGIONS BY WAVELET TRANSFORMATION

Wavelet transformation(WT) was widely used in many fields. We describe a WT system that predicts transmembrane helices in membrane proteins. In the WT system, we first compute propensity factors from 1160 membrane proteins as the training set. And then according to the propensity factors, we encoded protein sequences into propensity factors sequences. By using wavelet transformation as a low-pass filter, we obtained a smoothed sequence of propensity factor. Finally, we obtained the helix core from the peak value and expanded the helix core into a predicted helix region. The method was applied to 60 test membrane proteins with 228 HTM(helical transmembrane segment). Only 4 HTM were missed and the accuracy was about 92.85%.