airpart

airpart detects differential cell-type-specific allelic imbalance (AI) from single-cell RNA-sequencing (scRNA-seq) and other spatially- or temporally-resolved datasets to characterize cis-regulatory mechanisms in heterozygotes.

Key Features:

• Differential cell-type-specific allelic imbalance detection: Detects differential AI across cell types and states using scRNA-seq and spatially- or temporally-resolved data.
• Generalized Fused Lasso partitioning: Employs a Generalized Fused Lasso with Binomial likelihood to partition groups of cells based on AI signals while accounting for low-count single-cell data.
• Hierarchical Bayesian inference: Implements a hierarchical Bayesian model for statistical inference on allelic imbalance.
• Partitioning of genes and cells: Outputs discrete partitions that identify groups of genes and cells likely regulated by common cis-genetic mechanisms.
• Visualization and quality control: Produces visualizations and quality-control functions for examining single-cell allelic imbalance datasets.
• Performance in simulations and real data: Demonstrated lower RMSE of allelic ratio estimates in simulations and identified differential AI patterns and spatial/temporal trends in real datasets.

Scientific Applications:

• Cis-regulatory mechanism exploration: Reveals cis-regulatory contributions to allelic imbalance at single-cell resolution.
• Single-cell genomics research: Enables dissection of genetic interactions and regulatory networks in heterogeneous cell populations using scRNA-seq.
• Spatial and temporal genetic studies: Applies to spatially- and temporally-resolved datasets to study dynamics of gene expression and AI across biological contexts.

Methodology:

Uses a Generalized Fused Lasso with Binomial likelihood for partitioning groups of cells based on AI signals and a hierarchical Bayesian model for statistical inference on allelic imbalance.