我们知道，在基因芯片数据分析工作中，通常假定一个相同的profile中的基因往往具有功能相关性。但是，具有功能相关性的基因未必一定具有相同的profile，例如在一个代谢途径中的基因不一定有相同的表达坡面。下面这篇名为Transitive functional annotation by shortest-path analysis of gene expression data.的文章针对这一问题提出了自己的算法，并在酵母基因组中进行了研究。下面是文章的摘要：
Most techniques currently used to analyze microarray expression data sort genes according to common expression profiles. However, all genes within a biological pathway may not be similarly expressed. Researchers developed a computational technique that compares gene expression profiles in pairs, making it possible to assign the same biological pathway to two weakly correlated genes by connecting them with a string of transitive genes that exhibit related expression or function. Applying this analysis to yeast genome expression data revealed numerous genes related solely through transitive gene relationships. Researchers also assigned biological roles to about 5% of the predicted yeast genes whose function is currently unknown.
