T Cell Response Analysis Using RNA, Protein and TCR Profiling

Key Takeaways

• A single 10x Genomics 5' multi experiment yields three complementary layers of immune data - RNA, surface protein (CITE-seq), and TCR sequences - from the same cells, enabling cell identity, phenotype validation, and clonal tracking in one reproducible pipeline
• Reference-based annotation resolved 17 immune subtypes; 19 ADT protein markers independently validated every transcriptome-based assignment, demonstrating the power of orthogonal protein confirmation
• TCR repertoire analysis identified 4,852 unique clonotypes with antigen specificity predictions matching known CMV, EBV, and Influenza epitopes - establishing a high-resolution immune baseline directly applicable to immune monitoring and cell therapy development

Jump to section

Summary

This report presents an integrated multimodal analysis of ~10,000 human peripheral blood mononuclear cells (PBMCs) from a healthy donor, combining single-cell gene expression, 19 surface protein markers (CITE-seq), and paired T-cell receptor (TCR) sequencing from a single 10x Genomics 5' multi-experiment (10x Genomics). Together, these results establish a high-resolution immune baseline that teams can use to assess whether a patient's immune composition is within normal range, identify expanding or contracting T-cell populations in response to therapy, and prioritise clones for functional validation or cell therapy development.

• Cell type identification. Reference-based annotation identified 7 major lineages and 17 fine subtypes. T cells (53%) and monocytes (31%) dominate the PBMC compartment, with smaller populations of B cells (9.4%), ILC/NK (4.4%), dendritic cells, and pDC. Surface protein measurements from 19 ADT markers independently validate the transcriptome-based annotations.
• TCR repertoire. AIRR-compliant analysis processed 4,964 T cells with productive TCR rearrangements, yielding 4,852 unique clonotypes. The repertoire is highly diverse, consistent with a healthy background, with clonal expansion concentrated in CD8 effector/memory T cells (Tem/Temra).
• Antigen specificity. Database lookup matched 173 T cells to known epitopes, predominantly CMV (81 cells) and EBV (10 cells), consistent with prior viral exposure in a healthy adult.
• Workflow. The full pipeline runs reproducibly on the Sonrai platform with cloud-native orchestration for TCR analysis and a Python-based workflow for the CITE-seq component. All code, configuration, and outputs are version-controlled.

Introduction

Single-cell RNA-seq is sufficient to identify the major immune populations in peripheral blood, but multimodal assays provide a richer view of the immune state. Combining gene expression, cell-surface protein abundance (CITE-seq (Stoeckius et al. 2017)), and paired T-cell receptor (TCR) repertoires makes it possible to define cell identity, validate phenotypes with orthogonal protein measurements, and track clonal expansion within the same cells. This approach complements our AI-powered single-cell analysis workflow and our end-to-end RNA-seq workflow, extending the analytical toolkit available for translational immune monitoring.

This use case presents an end-to-end analysis of a public human PBMC dataset generated with the 10x Genomics 5' multi workflow:

• Cell Ranger multi for joint preprocessing of gene expression, feature barcode, and V(D)J libraries
• nf-core pipelines for AIRR-compliant TCR repertoire analysis (Ewels et al. 2020; Gabernet et al. 2024)
• Reference-based cell type annotation and Antibody-Derived Tags (ADT) protein validation
• TCR-epitope database lookup for antigen specificity prediction

The workflow is designed to run on AWS Batch with Spot instances and S3-backed storage, following the same cloud-native pattern used in the RNA-seq airway use case.

Methods

Data

The dataset is the public 10x Genomics Human PBMC from a Healthy Donor, 10k cells - multi (v2) (10x Genomics 2021).

Analysis Pipeline

The workflow is split into two coordinated branches: multimodal single-cell preprocessing and TCR repertoire analysis.

Step 1: Multimodal preprocessing with Cell Ranger multi

Step 2: TCR Repertoire Analysis

The TCR branch uses an nf-core/airrflow pipeline (Ewels et al. 2020; Gabernet et al. 2024) in assembled mode, taking the Cell Ranger AIRR rearrangement TSV as input for AIRR-compliant V(D)J annotation, clonotype inference, and repertoire statistics (CDR3 length distributions, V/J gene usage, clonal abundance).

Step 3: Downstream Multimodal Interpretation

Cloud execution: The workflow runs on AWS Batch with Spot instances and S3-backed storage. Public Cell Ranger outputs are staged to S3, a workflow orchestrator manages the TCR branch, and outputs are written back to S3 for local rendering - consistent with Sonrai's cloud-native pipeline infrastructure.

Results

Quality Control

After filtering (200-5,000 genes per cell, <15% mitochondrial reads), 10,213 cells were retained from the original 10,548 detected by Cell Ranger.

Cell Identity

Graph-based clustering identified 18 clusters. Reference-based annotation resolved these at two levels of granularity: 7 broad lineages (T cells, Monocytes, B cells, ILC, DC, pDC, Megakaryocytes/platelets) and 17 fine subtypes including naive and effector/memory subsets of both helper and cytotoxic T cells, MAIT cells, regulatory T cells, naive and memory B cells, classical and non-classical monocytes, DC2, pDC, and NK cells.

Protein-Validated Annotation

The panel includes 19 ADT markers spanning lineage (CD3, CD4, CD8a, CD14, CD19, CD56), activation (CD25, HLA-DR), differentiation (CD45RA, CD45RO), and checkpoint (PD-1, TIGIT) markers. Centered log ratio transformation (CLR)-normalised protein expression confirms the reference-based annotations at both cluster and cell type level. The checkpoint markers PD-1 and TIGIT show moderate expression concentrated in T cell subsets - a finding directly relevant to immune checkpoint target identification.

TCR Repertoire Structure

AIRR-compliant repertoire analysis identified 4,852 unique clonotypes across 4,964 T cells with matched RNA data. TCR detection rate was highest in conventional T cell subsets (>88% in Tcm/Naive helper and cytotoxic T cells, >92% in Tem/Temra cytotoxic T cells), 52% in MAIT cells (expected given their semi-invariant TCR), and low in non-T populations. The repertoire is highly diverse, consistent with a healthy donor: 4,775 singletons, 155 small clones (2-3 cells), and 34 expanded clones (4+ cells).

The five largest clonotypes are all CD8 T cells, consistent with antigen-driven clonal expansion of cytotoxic effectors:

Receptor Gene Usage and CDR3 Properties

TRBV gene usage shows a polyclonal distribution with no single dominant gene, consistent with a healthy background repertoire. CDR3 lengths follow the expected distribution for productive TRB rearrangements, centred around 13-15 amino acids.

Antigen Specificity Prediction

To explore potential antigen specificity, CDR3 beta sequences were queried against a curated database of experimentally validated TCR-epitope pairs. Exact CDR3 amino acid matching identified 173 cells (163 unique CDR3 sequences) with database hits, predominantly targeting common persistent virus epitopes:

• CMV (cytomegalovirus): 81 cells, including the immunodominant NLVPMVATV (pp65) and KLGGALQAK epitopes
• EBV (Epstein-Barr virus): 10 cells
• Influenza A: 5 cells (GILGFVFTL, M1 peptide)
• SARS-CoV-2: 2 cells
• HCV: 2 cells

This is consistent with a healthy adult donor who has likely been exposed to CMV and EBV, the two most common persistent herpesviruses in the human population.

Conclusion

This use case demonstrates a complete multimodal immune profiling workflow, from public 10x data to an integrated PBMC atlas with matched TCR clonotypes and antigen specificity predictions. The key results are:

• Cell identity: reference-based annotation resolves 17 immune subtypes from RNA, validated by 19 ADT surface protein markers
• TCR repertoire: AIRR-compliant analysis identifies 4,852 clonotypes; the repertoire is diverse with clonal expansion concentrated in CD8 effector T cells
• Antigen specificity: database matching reveals 173 T cells with known CMV, EBV, and Influenza reactivity, consistent with a healthy adult donor

The workflow connects transcriptome, surface phenotype, and receptor sequence in a single reproducible pipeline, directly applicable to immune monitoring and translational biomarker discovery. When applied to a disease cohort or treatment study rather than a single healthy donor, the same pipeline supports patient stratification by immune composition, identification of treatment-responsive clones, and prioritisation of candidates for cell therapy development. For a complementary multi-omics approach combining RNA-seq with proteomics and flow cytometry, see our MOFA+ autoimmune clinical trial case study.

Limitations

• Single healthy donor: findings are descriptive, not comparative. A multi-donor or disease-focused dataset would support stronger biological conclusions.
• Cell type annotation: reference-based annotation relies on training datasets that may not capture rare or atypical populations. ADT validation helps, but 3 of 19 markers are isotype controls (IgG1, IgG2a, IgG2b) with limited biological value.
• TCR-RNA barcode matching: 4,964 of 5,002 TCR cells matched RNA barcodes (99.2%), but the remaining cells were lost to QC filtering. The clonal threshold was set manually after automatic detection failed on a single-sample dataset.
• Specificity prediction: exact CDR3 matching is conservative and biased towards well-studied epitopes. Complementary approaches based on structural similarity clustering would provide broader coverage of convergent selection signals.

References

1. 10x Genomics. 2021. Human PBMC from a Healthy Donor, 10k Cells - Multi (V2).
2. Ewels, Philip A, et al. 2020. The Nf-Core Framework for Community-Curated Bioinformatics Pipelines. Nature Biotechnology 38: 276-78.
3. Gabernet, Guillaume, et al. 2024. Nf-Core/Airrflow: An Adaptive Immune Receptor Repertoire Analysis Workflow. Bioinformatics 40 (9): btae537.
4. Stoeckius, Marlon, et al. 2017. Simultaneous Epitope and Transcriptome Measurement in Single Cells. Nature Methods 14: 865-68.

No cost. No risk.

Want to apply this workflow to your own immune data?

Challenge Sonrai to find what others haven't. We'll run a multimodal immune profiling analysis on your dataset - at no cost.

Talk to our team

Sonrai acts as an extension of your team, combining scientific expertise, transparent and reproducible analysis, and a governed research environment to generate decision-ready evidence that can be revisited, reproduced, and built upon throughout the life of a programme.