OUTRIDER: A Statistical Method for Detecting Aberrantly Expressed Genes in RNA Sequencing Data.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• RNA sequencing (RNA-seq) is crucial for identifying rare disease causes by detecting aberrant gene expression.
• Current RNA-seq analysis methods lack statistical rigor or rely on subjective corrections for confounders.
• Accurate identification of outlier gene expression is essential for rare disease diagnostics.

Purpose of the Study:

• To introduce OUTRIDER, an algorithm for robustly detecting aberrant gene expression in RNA-seq data.
• To address limitations of existing methods by incorporating statistical significance and controlling for biological/technical variations.
• To provide an end-to-end solution for rare disease diagnostic platforms.

Main Methods:

• Utilizes an autoencoder to model gene expression expectations based on covariation.
• Models RNA-seq read counts using a negative binomial distribution with gene-specific dispersion.
• Identifies outliers as significant deviations from expected read counts, with automated model fitting and FDR-adjusted p-values.

Main Results:

• Simulations demonstrate the importance of controlling for covariation and using significance-based thresholds.
• OUTRIDER effectively identifies outlier read counts that deviate significantly from modeled expectations.
• The algorithm includes functionalities for gene filtering, outlier sample detection, and aberrant expression identification.

Conclusions:

• OUTRIDER offers a statistically sound and automated approach to detect aberrant gene expression in RNA-seq data.
• The algorithm enhances the molecular diagnosis of rare disorders by providing reliable identification of pathogenic gene expression.
• OUTRIDER is a valuable, open-source tool suitable for rare-disease diagnostic platforms.