1251 Pairing high-parameter spectral flow cytometry with CITE-seq using a novel automated artificial intelligence-based analysis platform to characterize immunophenotypes and cell states

<h3>Background</h3> Spectral flow cytometry and Cellular Indexing of Transcriptomes and Epitopes sequencing (CITE-seq) both enable high parameter single-cell surface protein analysis. Combining the two approaches allows for cost-efficient assessment of millions of cells from flow cytometry and deeper multi-omic profiling from the matched single-cell transcriptomic and proteomic data of CITE-seq. Analyzing flow cytometry and CITE-seq data often requires use of disparate analysis tools and platforms, and integrating results can be time and resource intensive. Here we apply Ozette Discovery™, an artificial intelligence-based analysis platform,¹ to analyze paired spectral flow cytometry and CITE-seq datasets generated from matched cryopreserved peripheral blood mononuclear cells (PBMC) isolated from healthy donor blood. <h3>Methods</h3> Cryopreserved PBMC from 8 healthy donors (table 1) were characterized by spectral flow cytometry using a 48-color pan-immune profiling panel. CITE-seq was performed on the same set of cryopreserved PBMC with 10X Genomics 3’ Gene Expression v3.1 and a 143-target antibody panel (table 2) with 6 isotype controls. The CITE-seq panel was composed of the Biolegend TotalSeq-B Universal Cocktail and a titrated 9-plex custom antibody mix to overlap the contents of the 48-color flow cytometry panel. Ozette Discovery was used to determine positive marker expression thresholds and identify immunophenotypes based on cell surface expression and detection frequency. <h3>Results</h3> After antibody derived tag counts were background-corrected and normalized, we detected 116 differentially expressed markers within the CITE-seq panel, of which 42 overlapped with the 48-color flow cytometry panel. Differentially expressed markers were used to identify major circulating immune cell lineages and expression of markers associated with specific cellular differentiation states, including activation (CD107a, CD27, CD28), exhaustion (CD39), and senescence (KLRG1), in both datasets. We compared cell phenotypic composition between the two technologies and assessed shifts in group frequencies, e.g. due to differences in sample preparation and quality control criteria. We found that immunophenotypes were differentially abundant between groups stratified by donor characteristics using both technologies. <h3>Conclusions</h3> Combining single-cell analytical methods allows researchers to leverage the advantages of each approach and better resolve immune cell lineages, subtypes, and cellular states in order to identify insights with translational impact. Ozette Discovery can be used to efficiently facilitate data analysis for multimodal studies to accelerate therapeutic target discovery and advancement of cancer immunotherapy research. <h3>Reference</h3> Allen D, Weaver M, Prokopchuk S, Lekschas F, Jiang M, Finak G, Greene E, McDavid A. Protein-based cell population discovery and annotation for CITE-seq data identifies cellular phenotypes associated with critical COVID-19 severity. <i>BioRxiv</i> 584720 [Preprint]. 2024 [cited 2024 June 24]. https://doi.org/10.1101/2024.03.14.584720.