Aims at the evaluation on the data of different Motif prediction algorithms, the optimization of traditional algorithms and the development of a new evaluation strategy. The method of our experiment was to develop the evaluation strategies of different prediction. Through the comparison and analysis of MEME algorithm and PKG algorithm, the sensitivity and specificity of Motifs in the same family and their corresponding ROC curves, the true Motifs was determined and then obtained the best Motifs model of the protein family. The results showed that the evaluation strategy can be applied to evaluate the results of Motifs prediction in protein family and to predict the other Motifs in protein family by the best motif database.%主要研究了如何评价蛋白质家族Motifs预测算法的预测结果,目的是在对传统的算法预测问题分析优化的基础上,制定新的评价策略.主要方法是通过对MEME算法和PKG算法预测结果的比较分析,计算同一家族中Motifs的敏感性和特异性并比较它们对应的ROC曲线,确定真实的Motifs,进而获得该蛋白质家族的最佳Motifs的模型.实验结果表明这种评价策略可用于算法对蛋白质家族Motifs预测结果的评价,还可利用确定的最佳Motifs搜索数据库来预测蛋白质家族中其他的Motifs.
展开▼
