[Paper Review] CEI: A Clonal Expansion Identifier for T-cell receptor clones following SARS-CoV-2 vaccination

Yunbei Pan, Christian Hofmann+5 more|arXiv (Cornell University)|Jan 28, 2026

vaccines and immunoinformatics approaches0 citations

TL;DR

CEI introduces a Python-based framework to detect vaccine-associated TCR clonotype expansions by jointly analyzing changes in clone abundance and log-fold change, using pre- and post-vaccination repertoires from COVID-naive individuals.

ABSTRACT

Each T cell typically carries a specific T-cell receptor (TCR) that determines its specificity against an epitope presented by the HLA complex on a target cell. Antigenic challenge triggers the expansion of reactive cells within a diverse pool of T cells with randomly generated receptors, a process that results in epitope-driven shifts of TCR frequencies over time. Here, we analyze the effects of SARS-CoV-2 vaccination on the TCR populations in peripheral blood drawn from seven COVID-naive individuals, before vaccines were widely available. To identify SARS-CoV-2 vaccine-associated TCR sequences among the $\sim 10^{5}-10^{6}$ TCR sequences sampled before and after vaccination, we develop statistical criteria to detect significant increases in abundance of positive TCR clones. Application of our statistical methods shows a robust identification of TCR sequences that respond to SARS-CoV-2 vaccination in vivo, illustrating the feasibility of quantifying the clone-specific dynamics of T-cell abundance changes following immunological perturbations.

Motivation & Objective

• Motivate the need to quantify clone-specific T-cell dynamics after SARS-CoV-2 vaccination in COVID-naive individuals.
• Develop a statistically principled method to detect vaccine-associated TCR clone expansions in large, replicate-limited repertoires.
• Provide a scalable Python package (CEI) for differential abundance analysis of TCRB clones across two conditions.
• Offer flexible approaches to control false discoveries and compare with existing methods under non-replicated data.

Proposed method

• Define two clone-level indicators: the absolute difference in normalized clone abundance d_i and the regularized log-fold change r_i with a Haldane-Anscombe correction.
• Model clone counts with a Poisson approximation and estimate variances using pooled proportions to compute Wald statistics Z_d,i and Z_r,i.
• Combine the two statistics into a joint score using a two-dimensional Mahalanobis distance with a trimmed medians-based baseline to identify expanded clones (CEI-Joint).
• Control false positives via multinomial permutation under the no-change null and apply Benjamini-Hochberg FDR correction.
• Offer an alternative aggregation of the two indicators using Aggregated Cauchy Association Test (ACAT) to produce a single p-value per clone (CEI-ACAT).
• Compare CEI-Joint and CEI-ACAT performance with heuristic Adap, and edgeR on the same data and across subjects and timepoints.

Experimental results

Research questions

• RQ1 Can vaccine-associated TCR clones be identified reliably in pre- and post-vaccination repertoires of COVID-naive individuals using replicate-limited data?
• RQ2 Do joint statistics that combine absolute abundance change and log-fold change improve detection of expanded TCR clones compared to single-metric approaches?
• RQ3 How do CEI-Joint and CEI-ACAT compare with existing methods (Adap, edgeR) in terms of sensitivity and false discovery control across subjects and timepoints?
• RQ4 Is the CEI framework generalizable to other perturbations and repertoire types beyond TCRB data?
• RQ5 What is the impact of different vaccination regimens (mRNA vs adenoviral) on the set of activated TCR clones identified by CEI?

Key findings

• CEI detects vaccine-associated TCR sequences with reproducible expansion signals in vivo, demonstrated across seven COVID-naive subjects.
• CEI-Joint generally identifies more expanded clones than CEI-ACAT, with substantial inter-subject and inter-visit heterogeneity in the number of positives.
• CEI-ACAT tends to be more conservative but can outperform CEI-Joint in specific subject/timepoint pairs.
• Compared with Adap and edgeR, CEI methods balance absolute differences and log-fold changes, often yielding different expansion profiles.
• Sensitivity analyses show the number of positives grows sharply with FDR threshold, indicating the methods’ tunable stringency.
• The framework accommodates replicate-limited data and can be applied to other two-condition abundance comparisons beyond TCR repertoires.