Immune receptor profiling enables tracking of individual T- or B-cell clones over time and across tissues, supporting studies of immune responses, cancer, and autoimmunity. Integrating receptor sequences with single-cell transcriptomes links clonotype to gene expression, revealing immune-cell functional states, differentiation pathways, and heterogeneity. Here we report a head-to-head evaluation of droplet-based microfluidics (10x Chromium Next GEM Single Cell 5' V(D)J v2 Dual Index) versus combinatorial barcoding (Parse Evercode™ paired TCR + WT v3).
Both platforms were benchmarked using biological and technical replicates of PBMCs from colorectal cancer patients after 14 days of in vitro differentiation into Cytokine-Induced Killer cells. Matched samples were processed on each technology and analyzed with vendor pipelines, applying gene-expression-guided filtering for consistent V(D)J cell calling. Parse whole-transcriptome FASTQs were downsampled to match 10x depth, while V(D)J libraries were evaluated at a targeted 5,000 reads per cell and assessed for sequencing saturation. Comparisons covered gene-expression QC, clustering, and repertoire metrics including T-cell proportions, clonotype overlap, CDR3 length distributions, and TRα/TRβ V–J usage.
Both platforms produced high-quality data and V(D)J libraries reached saturation. 10x called a higher proportion of T cells and recovered more TRα chains, whereas Parse showed lower ribosomal fractions and slightly lower ambient RNA estimates. The top 10 clonotypes were concordant, but overlap dropped to ~25% when comparing broader sets of top-ranked clonotypes. Across platforms, TRα recovery was lower than TRβ, and TRα V–J usage clustered by technology.
Overall, both technologies support integrated transcriptome–repertoire profiling, but platform effects can influence interpretation. Differences in cytoplasmic RNA capture, ambient RNA, T-cell yield, and sensitivity to lower-frequency clonotypes can shift estimates of repertoire diversity and clonotype sharing. Reporting sequencing depth and saturation, using gene-expression-guided V(D)J filtering, and checking chain-level recovery can improve comparability and reproducibility across studies.