基于Bootstrap方法的基因表达谱数据互信息估计研究

基因调控网络模型为深入理解生命本质提供了一个新的研究框架和平台。作为基因调控网络模型的其中一种，互信息关联网络模型使用熵和互信息描述基因和基因之间的关联。本文描述了用互信息度量基因表达相似性的方法，提出基于Bootstrap的互信息估计算法，并对产生的偏离现象提出了改进策略。实验结果表明，改进的互信息估计方法可以有效提高基因表达相似性估计的精确度。

Mutual Information Estimation Research of Gene Expression Data Based on Bootstrap Method

Genetic regulation network models provide a new research framework and platform for understanding the nature of life. As one of these models, mutual information relevance network model uses entropy and mutual information to describe the association of genes. This paper described the method to measure gene similarity in terms of expression profile using mutual information, presented a new method of mutual information estimation based on Bootstrap, which compensates for insufficient repeat samples. Furthermore, we analyzed the mechanism of the bias and brought up the strategy to reduce the bias. Our results indicated that the improved mutual information estimation method effectively raised the estimated accuracy.