sumrep

sumrep summarizes and compares adaptive immune receptor repertoires to quantify and characterize features of B-cell and T-cell receptor (IgH, IgL, IgK, TRA, TRB) AIRR-seq datasets for analysis and model validation.

Key Features:

• Summary statistics: Computes a wide array of summary statistics applicable at levels ranging from raw query reads to inferred phylogenies.
• Locus support: Explicitly analyzes IgH, IgL, IgK, TRA, and TRB loci.
• AIRR-seq compatibility: Operates on datasets generated through V(D)J recombination and subject to productivity-based filters and B-cell affinity maturation.
• Comparative plotting: Provides principled methods for plotting and comparing repertoire-level summaries and distributions.
• Covariate differentiation: Differentiates repertoires by covariates such as donor, timepoint, and cell type for both BCR and TCR data.
• Recombination feature analysis: Measures deletion and insertion length distributions from V(D)J recombination and compares their discriminative power to CDR3 amino acid composition.
• Model validation: Compares outputs of generative models to real repertoires, assessing recapitulation of gene usage, recombination statistics, and physiochemical properties.
• Implementation: Implemented as an R package.

Scientific Applications:

• Repertoire summarization and visualization: Summarizes and visualizes immune receptor diversity and distributional properties.
• Comparative repertoire analysis: Identifies differences across donors, timepoints, and cell types in BCR and TCR repertoires.
• V(D)J recombination characterization: Quantifies deletion and insertion length patterns and contrasts them with CDR3 amino acid composition.
• Generative model assessment: Validates and benchmarks generative models by comparing gene usage, recombination statistics, and physiochemical properties to real data.
• Immunological research and public-health analyses: Supports studies of adaptive immune diversity relevant to basic immunology and potential medical or public-health applications.

Methodology:

Implemented in R; computes summary statistics on AIRR-seq data from raw reads to inferred phylogenies, generates comparative plots, measures V(D)J deletion and insertion length distributions and CDR3 amino acid composition, and compares simulated/generative model outputs to real repertoires for gene usage, recombination statistics, and physiochemical property assessment.