Modelos lineares generalizados para análise de expressão gênica diferencial: biomarcadores sexo-específicos no Transtorno Depressivo Maior

Abstract

Generalized Linear Models (GLMs) play a crucial role in statistical analysis, offering a flexible approach for modeling data from various sources. An important application of GLMs is in modeling gene expression, allowing these models to handle the diversity of data distributions associated with gene expression, such as the Poisson or Negative Binomial distributions frequently observed in RNA sequencing data. GLMs accommodate biological variability by allowing modeling of non-constant variations in relation to the mean, providing a more accurate and realistic analysis. The inclusion of replicates is facilitated by GLMs, enabling the distinction between biological and technical variations, contributing to the solidity of conclusions. Additionally, the ability to incorporate covariates in GLMs is essential for controlling confounding factors, allowing a refined analysis that considers multiple biological and experimental aspects simultaneously. Here, we demonstrate an application of GLMs for the analysis of gene expression data using the approach implemented in the R/Bioconductor package edgeR. With this approach, we aim to identify gene expression changes in samples from individuals with Major Depressive Disorder (MDD) compared to samples from healthy individuals. For this purpose, we analyzed 263 RNA sequencing samples from post-mortem brain tissue from six brain regions: orbitofrontal cortex (OFC), dorsolateral prefrontal cortex (dlPFC), ventromedial prefrontal cortex (Cg25), anterior insula (aINS), nucleus accumbens (Nac), and ventral subiculum (Sub), in both sexes. This analysis resulted in the identification of 669 genes with altered expression in samples from individuals with MDD compared to samples from healthy individuals, referred to as differentially expressed genes. The set of differentially expressed genes constitutes a profile of the observed gene expression changes in MDD. This profile proved to be specific to each sex and each brain region considered in the analysis. Systematically, it was possible to identify groups of genes that interact physically and/or functionally. The biosynthesis of proteins, an important process related to neuronal metabolism, may be one of the biological processes impacted in MDD.