Detection of genes with differential expression dispersion unravels the role of autophagy in cancer progression

The majority of gene expression studies focus on the search for genes whose mean expression is different between two or more populations of samples in the so-called “differential expression analysis” approach. However, a difference in variance in gene expression may also be biologically and physiologically relevant. In the classical statistical model used to analyze RNA-sequencing (RNA-seq) data, the dispersion, which defines the variance, is only considered as a parameter to be estimated prior to identifying a difference in mean expression between conditions of interest. Here, we propose to evaluate four recently published methods, which detect differences in both the mean and dispersion in RNA-seq data. We thoroughly investigated the performance of these methods on simulated datasets and characterized parameter settings to reliably detect genes with a differential expression dispersion. We applied these methods to The Cancer Genome Atlas datasets. Interestingly, among the genes with an increased expression dispersion in tumors and without a change in mean expression, we identified some key cellular functions, most of which were related to catabolism and were overrepresented in most of the analyzed cancers. In particular, our results highlight autophagy, whose role in cancerogenesis is context-dependent, illustrating the potential of the differential dispersion approach to gain new insights into biological processes and to discover new biomarkers.Author summary: Gene expression is the process by which genetic information is translated into functional molecules. Transcription is the first step of this process, consisting of synthesizing messenger RNAs. During recent decades, genome-wide transcriptional profiling technologies have made it possible to assess the expression levels of thousands of genes in parallel in a variety of biological contexts. In statistical analyses, the expression of a gene is estimated by counting sequencing reads over a set of samples and is defined by two dimensions: mean and variance. The overwhelming majority of gene expression studies focus on identifying genes whose mean expression significantly changes when comparing samples of different conditions of interest to gain knowledge of biological processes. In this classical approach, the variance is usually considered only as a noise parameter to be estimated before assessing the mean expression. However, finely estimating the variance of expression may be biologically relevant since a modification of this parameter may reflect a change in gene expression regulation. Here, we propose to evaluate the performance of statistical methods that identify such differentially variant genes. We highlighted the potential of this approach by analyzing cancer datasets, thus identifying key cellular functions in tumor progression.