Gene Ontology based housekeeping gene selection for RNA-seq normalization

A Hypothesis Testing Based Method for Normalization and Differential Expression Analysis of RNA-Seq Data

“…In our method, we are supposed to have a set of housekeeping genes in priori, which could be reported in published studies or selected based on certain biological information, for example, the GO terms of the genes [14]. Assume we have m housekeeping genes in total and denote the set of housekeeping genes as H .…”

“…To evaluate the overall effectiveness of HTN, we also compare it with recent methods, including library size, TMM [10], IMM [11], Bull [13] and NHKS [14]. Both Bull and NHKS employ the information of housekeeping genes to determine the global scaling factor.…”

“…Chen et al [14] suggested to use a statistic called coefficient of variation, which measures the stability of gene expression, to select the most stable housekeeping genes. As in their studies [13, 14], we select a single and 15 housekeeping genes for Bull and NHKS, respectively. Note that for the sample with a single point, we can not calculate its coefficient of variation statistic.…”

“…Here, we consider the performance of the following methods: length-normalized count (Cloonan et al [17]), Poisson exact test [7] with library size, TMM [10], IMM [11], Bull [13], NHKS [14] and HTN normalization. The essence of virtual length [1] and RPKM [9] are the same as library size normalization and we do not compare them here.…”

“…Bullard et al [13] used a single housekeeping gene in normalization. Chen et al [14] proposed a method to select a subset of housekeeping genes by analyzing experimentally related GO terms and the stability of gene expression.…”