边排序贝叶斯网络结构学习算法应用于基因调控网络构建

提出一种基于多数据源融合思想的贝叶斯网络结构学习算法.该方法在现有贝叶斯网络结构学习算法的基础上,进行网络结构再学习,能有效处理不同数据源无法简单合并的问题.实验结果表明:在现有基因芯片数据节点数过多但数据量过少的前提下,该算法能有效提高建网精度;基于酿酒酵母细胞周期对不同实验条件下的表达数据进行融合,可以将正确率提高约12%.

Using Line Sorting Bayesian Structure Learning Algorithm to Get Gene Regulatory Network

This paper presents a new Bayesian network structure learning algorithm based on the idea of a fusion of multiple datasets. This algorithm can reconstruct the network with the help of existing Bayesian network structure learning algorithm, which can deal with the problem effectively that different datasets can’t be merged simply. With the microarray datasets of the Cerevisiae yeast cycle under different conditions, the algorithm can increase the correct rate by about 12%. Experimental results show that this algorithm canimprove the accuracy effectively even if there is few microarray data but too many genes.