Identification of differentially distributed gene expression and distinct sets of cancer-related genes identified by changes in mean and variability

Abstract

There is increasing evidence that changes in the variability or overall distribution of gene expression are important both in normal biology and in diseases, particularly cancer. Genes whose expression differs in variability or distribution without a difference in mean are ignored by traditional differential expression-based analyses. Using a Bayesian hierarchical model that provides tests for both differential variability and differential distribution for bulk RNA-seq data, we report here an investigation into differential variability and distribution in cancer. Analysis of eight paired tumour–normal datasets from The Cancer Genome Atlas confirms that differential variability and distribution analyses are able to identify cancer-related genes. We further demonstrate that differential variability identifies cancer-related genes that are missed by differential expression analysis, and that differential expression and differential variability identify functionally distinct sets of potentially cancer-related genes. These results suggest that differential variability analysis may provide insights into genetic aspects of cancer that would not be revealed by differential expression, and that differential distribution analysis may allow for more comprehensive identification of cancer-related genes than analyses based on changes in mean or variability alone.

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

“…Recently, four methods based on the NB distribution, MDSeq [39], DiPhiSeq [40], the analysis proposed by de Jong et al [41] and DiffDist [42], have been introduced to identify differences in both mean and dispersion in RNA-seq data within the same statistical framework.…”

“…DiffDist [42] implements a hierarchical Bayesian model which estimates mean and dispersion parameters based on log-normal priors whose location and scale parameters are modeled by normal and gamma hyperprior distributions, respectively. Posterior samples of the mean and dispersion parameters are obtained using an adpative Markov chain Monte Carlo algorithm.…”

“…DiffDist [42] was applied to all the simulated and TCGA datasets with the default values for all the parameters. The tail probabilities for both differential mean and differential dispersion PLOS COMPUTATIONAL BIOLOGY statistical were corrected by the Benjamini-Hochberg procedure [72] to control the FDR as recommended by the authors of DiffDist.…”

clrDV: A differential variability test for RNA-Seq data based on the skew-normal distribution

“…GAMLSS is based on the negative binomial 2 (NB2) model, whereby the mean and the variance are related quadratically as σ 2 = µ + ϕµ 2 . Recently, Roberts and others (2022) developed DiffDist, a hierarchical Bayesian model based on the NB2 model. In their work, gene expression variability is measured using the dispersion parameter ϕ , which is treated as a log-normal prior.…”

Mathematical model for the relationship between single-cell and bulk gene expression to clarify the interpretation of bulk gene expression data

“…DV can capture biological information about a target disease or trait. To date, studies have employed DV analysis of transcriptome data to provide biological insights about disease and aging [11] , [12] , [13] , [14] . For example, a strong relationship has been reported between variability in gene expression and a chronic lymphocytic leukemia subtype [14] .…”