Identifying stably expressed genes from multiple RNA-Seq data sets

摘要: We examined RNA-Seq data on 211 biological samples from 24 different Arabidopsis experiments carried out by labs. grouped the according to tissue types, and in each of groups, we identified genes that are stably expressed across samples, treatment conditions, experiments. fit a Poisson log-linear mixed-effect model read counts for gene decomposed total variance into between-sample, between-treatment between-experiment components. Identifying is useful count normalization differential expression analysis. The component analysis explore here first step towards understanding sources nature variation. When using numerical measure identify genes, outcome depends multiple factors: background sample set reference used normalization, technology measuring expression, specific stability used. Since (DE) measured relative frequencies, argue DE concept. advocate an explicit improve interpretability results, recommend common when analyzing avoid potential inconsistent conclusions.