简单的一致性方法预测蛋白质二级结构

Simple Consistency Method for Protein Secondary Structure Prediction

蛋白质二级结构预测对于我们了解蛋白质空间结构是至关重要的一步。文章提出了一种简单的二级结构预测方法,该方法采用多数投票法将现有的3种较好的二级结构预测方法的预测结果汇集形成一致性预测结果。从PDB数据库中随机选取近两年新测定结构的57条相似性小于30%的蛋白质,对该方法的预测结果进行测试,其Q3准确率比3种独立的方法提高了1.12—2.29%,相关系数及SOV准确率也有相应的提高。并且各项准确率均比同样采用一致性方法的Jpred二级结构预测程序准确率要高。这种预测方法虽然原理简单,但无须使用额外的参数,计算量小,易于实现,最重要的前提就是必须选用目前准确性比较出色的蛋白质二级结构预测方法。

Protein secondary structure prediction is an important step in understanding how proteins fold in three dimensions. To improve the prediction accuracy at a lower cost. This paper presents a method that combines existing three better secondary structure prediction methods by using majority voting to generate a result of consistency, 57 proteins, which had similarity less than 30% and were determined the 3D structure in the past few years, were randomly selected fi＇om the Protein Data Bank （PDB） to test prediction results of this secondary structure prediction, whose Q3 prediction accuracy, compared with the three independent methods, was increased by 1.12-2.29 per cent, and the correlation coefficients and SOV accuracy were also improved. All accuracy were higher than the secondary structure prediction of Jpred.This method is simple and easy to realize,which doesn＇t need to use extra parameters and only has small computational amount. In a word, a current method with excellent accuracy should be selected to predict protein secondary structure.