Differential gene expression analysis based on linear mixed model corrects false positive inflation for studying...

Area of Science:

• Genomics
• Bioinformatics
• Neuroscience

Background:

• Differential gene expression (DGE) analysis using RNA sequencing (RNA-Seq) is crucial for identifying genes related to specific traits.
• Existing DGE methods struggle with large RNA-Seq datasets and quantitative traits, often leading to inflated false positive rates.
• Linear mixed models (LMMs) are established in genetic association studies and offer a potential solution for DGE analysis.

Purpose of the Study:

• To adapt and apply a linear mixed model (LMM) for differential gene expression analysis in large-scale RNA sequencing data.
• To evaluate the performance of LMM in controlling false positive rates for quantitative Alzheimer's Disease (AD) traits.
• To identify genes differentially expressed in relation to AD traits and assess their replicability across different tissue types.

Main Methods:

• Employed a linear mixed model (LMM) for differential gene expression analysis of RNA sequencing data.
• Applied LMM to discovery RNA-Seq data from dorsolateral prefrontal cortex (DLPFC) tissue (n=632) associated with four continuous AD traits.
• Validated findings using quantile-quantile plots for p-value calibration and replication in additional RNA-Seq datasets from multiple tissues.

Main Results:

• LMM demonstrated well-calibrated false positive rates, unlike other methods that showed significant inflation.
• Identified 37 potentially significant differentially expressed genes in DLPFC for at least one AD trait.
• Replicated 17 of these significant genes in additional RNA-Seq data from DLPFC, supplemental motor area, spinal cord, and muscle tissues.

Conclusions:

• LMM provides a robust and well-calibrated approach for differential gene expression analysis in large RNA-Seq datasets, particularly for quantitative traits.
• The study identified novel candidate genes associated with Alzheimer's Disease traits.
• Findings suggest potential shared gene regulatory mechanisms underlying AD traits across various human tissues.