神经网络预测蛋白质二级结构的编码技术

Encoding of protein secondary structure prediction in neural network

考虑不同残基的物理化学性质 ,将其融入蛋白质二级结构的神经网络预测中 ,提出了一种新的归一化二进制编码技术 ,将各种残基不同的疏水值与体积大小归一化结果作为神经网络的二进制输入序列 .与其他蛋白质二级结构预测的方法进行了比较 ,结果表明 ,该方法能充分利用蛋白质的一级结构信息 ,获得了很好的预测效果 .

Based on orthogonal encoding mode in protein secondary structure prediction of neural network, a kind of encoding technology including a set of protein physical-chemical features was presented. Compared with other encoding methods, the method has better results using protein primary structure information.