Genome Biology recently published an article from Alicia Oshlack and colleagues in which they describe an approach for performing Gene Ontology analysis on RNA-seq data. RNA-seq is an emerging technology for monitoring gene expression levels by directly sequencing the mRNA molecules in a sample, and is likely to overtake microarrays as the technique of choice for gene expression profiling. Now, Genome Biology has published another innovative method, this time for normalizing RNA-seq data. This method is much needed and will be embraced by the genomics community as, until now, methods for normalizing RNA-seq data have often relied on tools that were based on those developed for microarray data.
A common approach for normalizing RNA-seq data has been to consider the expression of an individual gene relative to the global gene expression levels. In her latest paper, Oshlack, at the Walter and Eliza Hall Institute in Melbourne, Australia, shows that this is not always appropriate. In particular, if one tissue has a small number of genes that are significantly differentially expressed compared with another tissue then these can affect whether or not other genes in the sample are determined as being differentially expressed, often leading to implausible results. The paper demonstrates again the need for new statistical techniques to fully exploit the powerful RNA-seq technology, as well as providing a useful tool for doing this.