IgEvolution: clonal analysis of antibody repertoires

Abstract

Constructing antibody repertoires is an important error-correcting step in analyzing immunosequencing datasets that is important for reconstructing evolutionary (clonal) development of antibodies. However, the state-of-the-art repertoire construction tools typically miss low-abundance antibodies that often represent internal nodes in clonal trees and are crucially important for clonal tree reconstruction. Thus, although repertoire construction is a prerequisite for follow up clonal tree reconstruction, the existing repertoire reconstruction algorithms are not well suited for this task. Since clonal analysis has the potential to reveal errors in the constructed repertoires and contribute to constructing more accurate repertoires, we advocate a tree-guided construction of antibody repertoires that combines error correction and clonal reconstruction as interconnected (rather than independent) tasks. We developed the IgEvolution algorithm for simultaneous repertoire and clonal tree reconstruction and applied it for analyzing multiple immunosequencing datasets representing antigen-specific immune responses. We demonstrate that analysis of clonal trees reveals highly mutable positions that correlate with antigen-binding sites and light-chain contacts in crystallized antibody-antigen complexes. We further demonstrate that this analysis leads to a new approach for identifying complementarity determining regions (CDRs) in antibodies.

V(DD)J recombination is an important and evolutionarily conserved mechanism for generating antibodies with unusually long CDR3s

“…We classify a clonal tree (lineage) as large if it contains at least minLineageSize sequences (the default value minLineageSize = 30). IgEvolution (Safonova and Pevzner 2019b) (Safonova and Pevzner 2019b) applied to all reads from these lineages. These clonal lineages originated from a V(DD)J recombination that resulted in long CDR3s of length 72 and 78 nt, respectively.…”

The influence of the phylogenetic inference pipeline on murine antibody repertoire sequencing data following viral infection

“…Assigning sequences from bulk Ig-seq data is typically one of the first steps in reconstructing B cell clonal lineages (Yermanos et al, 2018). Although multiple tools specifically tailored to clonal lineage assignment now exist (Ralph and Matsen IV, 2016;Briney et al, 2016;Schramm et al, 2016;Safonova and Pevzner, 2019), a vast number of studies have performed some variation of first aligning the recovered antibody sequences to the reference germline sequences and subsequently clustering based on germline gene usage, edit distance sequence homology and/or CDR3 length (Stern et al, 2014;Doria-Rose et al, 2014;Bhiman et al, 2015;Tsioris et al, 2015). While different germline aligner tools have been previously compared (Marcou et al, 2018), how the aligner tools impact the repertoire fingerprint remains less characterized.…”

Scalable Models of Antibody Evolution and Benchmarking of Clonal Tree Reconstruction Methods

“…Nevertheless, the impact of the error on various methods and the overall accuracy should be tested in future work. Similarly, the efficacy of methods that simultaneously filter errors and build clonal trees (e.g., Safonova and Pevzner, 2019;Lee et al, 2017) should be subject of future research.…”

“…Finally, our 5-mer mutation model, while based on the empirical model of Yaari et al (2013), still fails to capture some of the complexities of the real antibody evolution. For example, we concentrated substitutions on the CDR region, but other regions are known to also accumulate mutations (Safonova and Pevzner, 2019;Kirik et al, 2017;Ovchinnikov et al, 2018). Other B cell specific models (e.g., Elhanati et al, 2015) including those that seek to tease out the effects of selection from background mutations (e.g., McCoy et al, 2015) and per-position mutability models (Kepler et al, 2014) can be incorporated in the future.…”

“…Memory B cells can be repeatedly activated and subjected to the AM, while plasma B cells can secrete massive levels of neutralizing antibodies. Recent studies show that CDRs, which include the binding sites, accumulate more SHMs compared to FRs (Hsiao et al, 2019;Safonova and Pevzner, 2019).…”