【摘要】 基于Z标度法对C/18家族几丁质酶的特征序列PS01095和PS00232进行数字转换,将得到的数据集采用逐步回归方法回归预测,构建了几丁质酶特征序列与其最适pH间关系的数学模型。当模型的相关系数为R=0.964,显著水平P<0.001,得到了最佳的预测效果。模型对pH值拟合的平均绝对百分比误差为0.05%,同时具有良好的预测效果,预测的平均绝对误差为0.26个pH单位,比基于几丁质酶氨基酸组成的支持向量机模型更好。更多还原

【Abstract】 The characteristic sequences PS01095 and PS00232 of C/18 family chitinase were translated into Z-scales, which then were used as input data to establish a prediction model for optimum pH based on stepwise regression. When the correlation coefficient R=0.964 and signifcance level P<0.001, the prediction pHs fitted the experimental pHs very well and the MAPEs (mean absolute percent error) was 0.05 %. At the same time, the prediction pHs fitted the experimental pHs well and the MAE (mean absolute error) was 0.26 pH unit. The present model was better than that based on support vector machines, whose input data was the amino acid composition of whole chitinase.更多还原