基于图聚类的蛋白质功能预测方法

Predicting Protein Function Based on Graph Clustering

利用蛋白质序列的循环关系,采用循环匹配算法对数据进行预处理,得到相关联蛋白质数据集,再利用该数据集构造蛋白质的网络图,在此基础上采用图聚类算法,对待预测的蛋白质相关的各个蛋白聚类,并进行子群分割,对各个子群采用z值进一步计算并得出作为预测结果的蛋白质功能.经实验,该方法与其它最新方法相比较,预测结果的最终衡量指标F1-measure具有明显的提升.

This ccording to the circular relationship of protein, and binding protein domain representation, then it constructs the protein network diagram. Based on this network, it uses a graph clustering method to divide the proteins which associated with the unknown proteins into serval subgroups, then calculate the z score to make the further decision to filter the result. Compared to the recent results of other methods, its predicted result of the final measure F1-measure has the obvious promotion.