Background The use of single cell sequencing for deep immune monitoring in discovery and clinical settings is accelerating. Cellular indexing of transcriptomes and epitopes (CITE-seq) is a version of single cell sequencing that generates highly multiplexed protein and gene expression readouts. Dimensional reduction and clustering are the standard methods used to define immune cell populations in CITE-seq analyses. In contrast, hierarchical gating has been the approach favored by Immunologists for decades to identify cell populations in flow cytometric single cell datasets. Here we evaluated the use of hierarchical gating to identify immune cell types in CITE-seq data.

Methods CITE-seq data was generated on replicate healthy donor PBMC samples using 10X Genomics workflows established at CellCarta. A bioinformatics pipeline was developed to generate 1,000+ parameter flow cytometry standard (FCS) files that included individual gene and protein features along with calculated features such as gene signature scores. Immune cell populations were identified by clustering, hierarchical gating, or a hybrid approach where hierarchical gates were adjusted based on insights from clustering. The variation of assay readouts was evaluated for each cell type identification method to understand their impact on assay repeatability.

Results Hierarchical gating refined by clustering, which we call hybrid hierarchical gating, defined the broadest and most uniform set of immune cell types in our CITE-seq dataset. Clustering alone did not identify every immune population we would typically gate in a cytometry experiment; however, it did help identify distinct cell subsets within our gated populations that our hierarchy initially missed. Cell populations defined by the hybrid approach tended to be more uniform with assay readouts that showed lower coefficients of variation.

Conclusions Hybrid hierarchical gating identified immunologically relevant cell types, helped ensure key subpopulations were not missed, and improved the repeatability of assay readouts from our CITE-seq dataset. By anchoring CITE-seq analysis in hierarchical gating, which is familiar to Immunologists, we show how complex sequencing data can be analyzed more intuitively than many realize today.