Biology subjects

Basavaraju, Y.

Publications and source records attributed to Basavaraju, Y..

3 recordsLinked to original sources

bioRxiv2026

Exploring noise in TCR specificity data through supervised learning

Accurate prediction of T cell receptor (TCR) specificity can potentially accelerate development of novel immunotherapies, yet public databases used to train these models contain substantial label noise. This uncertainty complicates model evaluation and limits the reliability of predicted TCR-pMHC interactions. In this work, we present a simple supervised machine learning denoising scheme to identify low-quality labels in TCR specificity data. We evaluated the framework using the TCRvdb validation dataset and further corroborated it via structural modeling with AlphaFold 3. Here, our denoising approach successfully identified non-validated TCRs, achieving an average MCC of 0.47 for filtering non-validated examples while retaining high sensitivity for validated examples. Applying our denoising approach to the complete NetTCR training dataset, we observed that data associated with known noisy attributes were significantly enriched in the filtered subset. Furthermore, TCR-pMHC complexes identified as noise consistently produced lower AlphaFold 3 confidence scores, providing independent evidence of misannotation. Experimental validation of 30 selected TCRs further supported the approach, with 80% of TCRs predicted as noise failing to demonstrate functional reactivity. Our findings demonstrate that current performance metrics may be obscured by dataset inaccuracies and provide a method for refining training data in the absence of gold-standard benchmarks.

bioinformatics↗

bioRxiv2026

KIR Expression Defines a Transcriptionally Distinct CD8+ T-cell Population that Confounds Antigen-Specific T-cell Detection

HLA-C-restricted CD8+ T-cell responses remain poorly defined compared with HLA-A and HLA-B responses, despite the central role of HLA-C as both a peptide-presenting molecule and a ligand for killer-cell immunoglobulin-like receptors (KIRs). This dual function creates a major challenge for pHLA multimer-based antigen-specific T-cell analysis, as HLA-C multimers may bind CD8+ T-cells through either antigen-specific TCRs or KIRs. Here, we show that KIR expression on CD8+ T-cells drives TCR-independent binding to pHLA-C multimers, leading to overestimation of HLA-C-restricted antigen-specific T-cell responses. Using HLA-mismatched donor settings and cancer neoantigen pHLA-C multimers, we demonstrate that apparent multimer-positive CD8+ T-cells can arise from KIR-mediated recognition rather than cognate TCR specificity. We further show that KIR blockade before pHLA-C multimer staining removes non-TCR-specific binding, and applying this strategy to large-scale analysis of CMV, EBV, and SARS-CoV-2 antigens, we identify TCR-specific HLA-C-restricted CD8+ T-cell populations. Importantly, KIR-mediated pHLA-C binding is peptide-selective rather than uniform across all peptide-HLA-C complexes. Single-cell transcriptomic, phenotypic, and clonotypic analyses further demonstrate that KIR-mediated and TCR-mediated pHLA-C multimer binding define transcriptionally and clonally distinct CD8+ T-cell populations. Together, these findings establish KIR-mediated pHLA-C recognition as a major confounder in HLA-C multimer-based T-cell analysis. By separating KIR- from TCR-mediated binding, our approach provides a practical framework for accurate discovery and characterization of HLA-C-restricted antigen-specific CD8+ T-cells across viral infection and cancer.

immunology↗

bioRxiv2026

DNA-barcode labelled MHCII multimers for detection of antigen-specific CD4 T cells across large libraries of epitopes

The role of antigen-specific T cells responding to antigen is a topic of intense studies, and critical for mechanistic insight of diseases and development of therapeutic strategies. Methods for broad-scale detection of antigen-specific CD4 T cells are lacking, while such methods have demonstrated great value in exploring CD8 T cell response in health and disease. Furthermore, major histocompatibility complex II (MHCII) assays are technically challenging due to high HLA diversity, lower binding affinities, low frequencies of ex vivo antigen-specific CD4 T cells and several bottlenecks in production and peptide exchange of MHCII monomers. Here we use peptide-loaded MHCII (pMHCII) proteins multimerized on a barcode- and fluorophore-labelled dextran backbone to provide a method for the detection of peptide-specific CD4 T cells by using a large display of MHCII-associated peptides. We have established a protocol for MHCII production and peptide-exchange suitable for the generation of large libraries of peptide-MHCII complexes. We validate the use of such pMHCII complexes in the form of barcode-labelled MHCII multimers to detect antigen-specific CD4 T cells. We demonstrate that we can identify antigen specific CD4 T cells, using these DNA barcoded peptide-MHCII multimer. The multimer bound CD4 T cells were selected based on the fluorochrome signal, and the co-attached DNA barcodes were hereafter amplified and used to identify the peptide-MHCII response/binding. In cases where the peptide-specific CD4 T cells frequencies are very low, we expanded the cell population with peptide-pools and in the presence of IL2. The given CD4 T cell populations hereby reach a cell number allowing for the DNA-barcoded pMHCII multimers to detect responses otherwise missed out. Applying this technology, we utilized a panel of 150 peptides derived from human cytomegalo virus (CMV), Epstein barr virus (EBV), Influenza (Flu), SARS CoV 2 and SARS CoV1, Hepatitis B virus (HBV), and Hepatitis C virus (HCV) loaded onto HLA-DRB1*01:01 and DRB1*04:01 to screen peripheral blood mononuclear cells (PBMC). We assessed ex vivo responses in 16 participants with HCV infection, and successfully detected naturally occurring viral-specific CD4 T cells at frequencies as low as 0.004% of total CD4 T cells. The low-frequency responses, identified via the barcode screen, were rigorously validated using individual fluorophore-labelled tetramer staining after a peptide-driven expansion in 15 participants. Furthermore, we assessed the recognition of novel HCV epitopes in 11 additional participants. Through this, we identified a total of 12 distinct HCV epitopes, including 9 that have not been previously utilized in assays to detect CD4 T cells. Overall, this barcoded-multimer platform provides a powerful tool for the large-scale discovery of class II epitopes and the broad profiling of CD4 T cell specificities. This method will allow for in-depth analyses of immune interactions, provide a better understanding of the antigen-driven associations between CD4 and CD8 T cell responses, and help dissect the complexities of CD4 T cell protection in HCV infection.

immunology↗