Reconstructing antibody repertoires from error-prone immunosequencing datasets

TL;DR

This paper introduces IgReC, a novel algorithm for reconstructing antibody repertoires from error-prone immunosequencing data, and a new framework for evaluating their accuracy, showing it can outperform or match barcoding-based methods.

Contribution

The paper presents IgReC, a new computational method for antibody repertoire reconstruction that reduces reliance on barcoding technology, with a comprehensive benchmarking framework.

Findings

IgReC achieves high accuracy in repertoire reconstruction.

Reconstructed repertoires from IgReC outperform existing tools.

Barcoded and non-barcoded reconstructions by IgReC are comparable.

Abstract

Transforming error-prone immunosequencing datasets into antibody repertoires is a fundamental problem in immunogenomics, and a prerequisite for studies of immune responses. Although various repertoire reconstruction algorithms were released in the last three years, it remains unclear how to benchmark them and how to assess the accuracy of the reconstructed repertoires. We describe a novel IgReC algorithm for constructing antibody repertoires from high-throughput immunosequencing datasets and a new framework for assessing the quality of reconstructed repertoires. Benchmarking IgReC against the existing antibody repertoire reconstruction tools has demonstrated that it results in highly accurate repertoire reconstructions. Surprisingly, antibody repertoires constructed by IgReC from barcoded immunosequencing datasets in blind mode (without using unique molecular identifiers information) improved upon the repertoires constructed by the state-of-the-art tools that use barcoding. This finding suggests that IgReC may alleviate the need to generate repertoires using the barcoding technology (the workhorse of current immunogenomics efforts) because our computational approach to error correction of immunosequencing data ends up being nearly as powerful as the experimental approach based on barcoding.