SIMCA法用于从非同源蛋白一级序列预测其结构类

SIMCA method and application in prediction of protein secondary structural classes

蛋白质结构类的正确识别对于其三级结构预测具有十分重要的意义,有必要引入先进的算法提高预测精度。使用SIM-CA 法处理氨基酸组成、自相关系数提取的特征参数以及氨基酸对含量,进行了蛋白质结构类的预测。采用Miyazawa 和Jerni-gan 的疏水值时,All-α、All-β、αβ类的自检验的精度为89%、91%、89%,它检验的精度分别为74%、87%、91%;引入氨基酸对含量后,All-α、All-β、αβ类自检验精度为86%、89%、90%,它检验的精度为77%、88%、93%。SIMCA 的预测结果好于Bayes-ian 识别函数法,氨基酸对的引入可以提高预测精度。

Protein structural classes prediction plays important role in three-dimensional structure prediction,advanced algorithmshould be introduced.Amino acid compositions,the auto correlation function and amino acids pairs compositions are used to describesequence information encoding in protein,then SIMCA method is introduced to predict the secondary classes of protein.Self-consisten-cy prediction accuracy rates of All-α,All-β,αβ are 89%,91%,89%,and cross-validation accuracy rates are 74%,87%,91%when Miyazawa and Jernigan's index is used.After amino acids pairs compositions are introduced,self-consistency prediction accuracyrates of All-α,All-β,αβare 86%,89%,90%,and cross-validation accuracy rates are 77%,88%,93%.The result is better thanthe method based bayesian discriminant function and the introduction of amino acids pairs compositions can improve prediction accura-cy.