Department of Sciences, Dalian Nationalities University, Dalian, Liaoning 116600, PR China.
College of Life Sciences, Zhejiang Sci-Tech University, Hangzhou, Zhejiang 310018, PR China.
Biochimie. 2014 Feb;97:60-5. doi: 10.1016/j.biochi.2013.09.013. Epub 2013 Sep 22.
Knowledge of protein secondary structural classes plays an important role in understanding protein folding patterns. In this paper, 25 features based on position-specific scoring matrices are selected to reflect evolutionary information. In combination with other 11 rational features based on predicted protein secondary structure sequences proposed by the previous researchers, a 36-dimensional representation feature vector is presented to predict protein secondary structural classes for low-similarity sequences. ASTRALtraining dataset is used to train and design our method, other three low-similarity datasets ASTRALtest, 25PDB and 1189 are used to test the proposed method. Comparisons with other methods show that our method is effective to predict protein secondary structural classes. Stand alone version of the proposed method (PSSS-PSSM) is written in MATLAB language and it can be downloaded from http://letsgob.com/bioinfo_PSSS_PSSM/.
蛋白质二级结构类别的知识对于理解蛋白质折叠模式起着重要作用。在本文中，选择了 25 个基于位置特异性评分矩阵的特征来反映进化信息。结合前人提出的基于预测蛋白质二级结构序列的其他 11 个合理特征，提出了一个 36 维的表示特征向量，用于预测低相似度序列的蛋白质二级结构类别。ASTRALtraining 数据集用于训练和设计我们的方法，其他三个低相似度数据集 ASTRALtest、25PDB 和 1189 用于测试所提出的方法。与其他方法的比较表明，我们的方法能够有效地预测蛋白质二级结构类别。所提出的方法的独立版本（PSSS-PSSM）是用 MATLAB 语言编写的，可以从 http://letsgob.com/bioinfo_PSSS_PSSM/ 下载。
