sumrep: a summary statistic framework for immune receptor repertoire comparison and model validation

Abstract

The adaptive immune system generates an incredible diversity of antigen receptors for B and T cells to keep dangerous pathogens at bay. The DNA sequences coding for these receptors arise by a complex recombination process followed by a series of productivity-based filters, as well as affinity maturation for B cells, giving considerable diversity to the circulating pool of receptor sequences. Although these datasets hold considerable promise for medical and public health applications, the complex structure of the resulting adaptive immune receptor repertoire sequencing (AIRR-seq) datasets makes analysis difficult. In this paper we introduce sumrep, an R package that efficiently performs a wide variety of repertoire summaries and comparisons, and show how sumrep can be used to perform model validation. We find that summaries vary in their ability to differentiate between datasets, although many are able to distinguish between covariates such as donor, timepoint, and cell type for BCR and TCR repertoires. We show that deletion and insertion lengths resulting from V(D)J recombination tend to be more discriminative characterizations of a repertoire than summaries that describe the amino acid composition of the CDR3 region. We also find that state-of-the-art generative models excel at recapitulating gene usage and recombination statistics in a given experimental repertoire, but struggle to capture many physiochemical properties of real repertoires.

Deep generative models for T cell receptor protein sequences

“…We next sought to quantify the similarity of model-generated sequences to real sequences, for each of the three models in consideration. To accomplish this task, we used the sumrep package (Olson et al, 2019) (https://github.com/matsengrp/sumrep/), a collaborative effort of the AIRR (Breden et al, 2017; Rubelt et al, 2017) software working group. This package calculates many summary statistics on immune receptor sequence repertoires and provides functions for comparing these summaries.…”

Benchmarking of T cell receptor repertoire profiling methods reveals large systematic biases

“…[1]Two such approaches have been proposed for specific clone detection in Minimal Residual Diseases 45,46 as well as for the BCR, but not TCR, repertoire 47 , still at a very low diversity level. The construction of such gold standard repertoires is currently very costly and remains a major challenge that the Adaptive Immune Receptor Repertoire Community (AIRR-C) 48 , engaged in AIRR-seq standardization [49][50][51] , may tackle in the future. Finally, in this study some data were pre-processed using proprietary (mPCR-1, mPCR-3) or published 30,52 (RACE-1_U and RACE-2_U) tools and then aligned and error-corrected using MiXCR (v2.1.10) 37 .…”

“…For RACE-1 and RACE-2, UMI pre-processing was performed following protocols published elsewhere 29,30 . FASTQ and FASTA files were then processed for TRB and TRA sequence annotation using the MiXCR software (v2.1.10) with RNA-Seq parameters (-p rna-seq -s hsa) 50 .…”

TCR+/BCR+ dual-expressing cells and their associated public BCR clonotype are not enriched in type 1 diabetes

“…While these efforts raise concerns over the validity of DE cells as a biomarker or mechanism of T1D pathogenesis, our data provide an opportunity for further discovery. The data can be used to determine whether potential immune cell or immune repertoire motifs are associated with T1D, to compare against libraries of antibodies with known auto-specificities (Seay et al, 2016) as well as for sample normalization, motif discovery, and disease-specific immune subset analysis (Olson et al, 2019;Miho et al, 2019). Another interesting and under-explored aspect of our data is the analysis of the Aab + control subjects, who have broken tolerance to a subset of T1D-associated autoantigens (Battaglia et al, 2020).…”