摘要
本文使用人工神经元网络预测蛋白质分子主链的二面角,通过分别学习所有残基,成规则二级结构的残基,无规线团的二面角,结果表明由蛋白质分子的一级序列出发预测主链二面角,对于α螺旋成功率最高,β折叠次之,而无规线团的二面角和其一级序列相关很小,故难以预测.
A method for predicting the main-chain dihedral angles of protein by using artificial neural network methods is presented. By training the perceptron with the known dihedral angles in X-ray crystallographic structures, the perceptron can then be used to predict the dihedral angles based on the primary sequence. We have also trained the perceptron with only dihedral angles of residues in β-helix and β-sheet or with only those in random coil. The results indicate that the predicting performance is best for the dihedral angles in α-helix and then those in the β-sheet. The dihedral angles in the random coil have little correlations with its primary sequence and therefore are difficult to predict.
出处
《生物物理学报》
CAS
CSCD
北大核心
1991年第2期157-160,共4页
Acta Biophysica Sinica
基金
国家863高技术项目资助课题
