COTAN

COTAN analyzes gene co-expression in single-cell RNA sequencing (scRNA-seq) datasets by modeling zero unique molecular identifier (UMI) counts to detect correlated and anti-correlated gene pairs and to score differential expression at the single-cell level.

Key Features:

• Zero-UMI distribution modeling: Examines the distribution of zero UMI counts instead of relying only on positive counts to address sparsity and noise in scRNA-seq data.
• Generalized contingency tables: Uses a generalized contingency tables framework to assess co-expression between gene pairs within single cells.
• Novel correlation index: Introduces a correlation index with an approximate p-value for statistical testing of independence between genes.
• Global differentiation index: Computes a global differentiation index that scores individual genes for differential expression at the single-cell level.
• Co-expression network analysis: Provides methods for constructing, visualizing, and clustering genes based on co-expression patterns to reveal gene interaction modules.

Scientific Applications:

• Gene-pair independence testing: Detects correlated and anti-correlated expression patterns between gene pairs in scRNA-seq data.
• Differential expression scoring: Identifies differentially expressed genes and potential cell-identity markers using the global differentiation index.
• Gene interaction and module discovery: Elucidates gene interactions and clusters genes into co-expression modules via network analysis.
• Developmental biology case study: Applied to neural development scRNA-seq datasets to investigate gene co-expression dynamics.

Methodology:

Models zero UMI count distributions, applies generalized contingency tables for gene-pair co-expression analysis, computes a novel correlation index with approximate p-values for independence testing, calculates a global differentiation index for genes, and performs correlation network analysis and gene clustering.