基于NC-SVM的酵母蛋白功能预测研究

Prediction Study of the Protein Function for Yeast Protein Based on NC-SVM

蛋白质功能的预测是当前生物信息技术的一个重要研究方向,它广泛应用于分析新发现蛋白质的功能及人工合成的蛋白质功能。利用酵母蛋白相互作用数据,首先针对传统的基于支持向量机和基于邻近投票法的功能预测法分析了它们各自的优劣性,然后研究出结合2种预测方法的优势的NC-SVM法对酵母蛋白的功能进行预测,采用科学的评价方法,通过反复试验并对结果进行分析,发现这种新方法的预测结果优于单独使用前2种,而且具有一定的稳定性。

In the proteomics formed in the post-genomic era,one of the most important areas of bioinformatics is prediction of protein function using for analyzing the protein of new protein or the artificial protein.Based on database of the protein-protein interaction for yeast protein,the merits and the demerits both of the Support Vector Machine and the Neighbor Counting method were analyzed,then combining the two methods,a new method called NC-SVM was engendered.Through the experiments,with the scientific evaluation method,it＇s conclued that the new one was better than the ones before,and showed certain stability.