scLANE

scLANE builds flexible, interpretable models of single cell gene expression over pseudotime or latent time using truncated power basis spline models.

Key Features:

• Multiple modeling architectures: Supports Negative-binomial GLMs, GEEs, and GLMMs for trajectory differential expression analysis.
• Empirical basis spline selection: Handles nonlinearity with basis splines chosen empirically for each gene.
• Model comparison: Provides functionality to compare different fitted models.
• Dynamic gene clustering: Clusters genes based on their dynamic expression patterns over pseudotime.
• Smoothed counts generation: Generates smoothed count values from fitted models.
• Gene set enrichment testing: Performs enrichment testing on sets of genes.
• Visualization: Includes visualization capabilities for results.

Scientific Applications:

• Trajectory differential expression: Identifies genes with expression changes significantly associated with progression through biological trajectories in single-cell RNA-seq data.
• Pseudotime analysis: Models gene expression dynamics downstream of pseudotime estimation methods.
• RNA velocity analysis: Provides biological information when used downstream of RNA velocity estimation methods.
• Complex experimental designs: Enables reliable trajectory testing through extensions to estimating equations and mixed models.

Methodology:

scLANE uses truncated power basis spline models with empirical selection of basis splines per gene, implemented within an interpretable generalized linear model framework with extensions to generalized estimating equations and mixed models.