MEB

MEB implements a normalization-invariant minimum enclosing ball method to identify differentially expressed genes in the same or different species from RNA-seq and scRNA-seq data.

Key Features:

• Scaling-free minimum enclosing ball (SFMEB) model: Builds a model to cover non-DE genes in feature space based on their similarities, treating DE genes as outliers rejected outside the ball.
• scMEB extension: Extends the SFMEB method to single-cell RNA-seq data for identifying differentially expressed genes in datasets with two or more potential cell types or unknown labels.
• Cross-species analysis: Supports differential expression gene identification across different species.
• Normalization invariance: The method is invariant to normalization procedures applied to the data.

Scientific Applications:

• Bulk RNA-seq differential expression analysis: Detects differentially expressed genes from RNA-seq data by modeling non-DE genes as inliers within a minimum enclosing ball and DE genes as outliers.
• Single-cell RNA-seq differential expression analysis: Identifies differentially expressed genes in scRNA-seq datasets with multiple potential cell types or unknown cell labels using the scMEB extension.
• Cross-species gene expression comparison: Enables identification of differentially expressed genes when comparing samples from the same or different species.

Methodology:

The scaling-free minimum enclosing ball (SFMEB) model constructs a ball in feature space to cover non-DE genes based on their shared features, with DE genes detected as outliers falling outside the ball. For single-cell data, the scMEB method extends this approach to handle datasets with two or more potential cell types or unknown labels.