UCSF

Abstract: Cellular Indexing of Transcriptomes and Epitopes (CITE-Seq) enriches single cell transcriptomic insights by incorporating information about the cell surface phenotype through the application of oligonucleotide-tagged monoclonal antibodies. Similar to observations in flow cytometry, surface protein quantification in CITE-Seq introduces technical noise, potentially obscuring underlying biological processes and giving rise to misleading results. This noise originates from ambient antibodies within the reaction compartment (such as droplets or wells) as well as non-specific binding. Here, we present ThresholdR, an R-based automated tool, to reliably and systematically find the threshold for each antibody under the experimental conditions used in each experiment. We show that ThresholdR helps identification of cell types and enhances the presentation of biological differences across cell populations. Flow cytometry validation confirmed the proportions of cell populations per patient identified by ThresholdR, demonstrating strong and statistically significant correlations. Furthermore, our findings reveal that cell type-specific noise stems from the non-specific binding of antibodies, with larger cells, such as monocytes, exhibiting higher levels of noise compared to other cell types. We systematically assess the performance of ThresholdR across diverse datasets and platforms, underscoring the agreement between individual cell data and well-established surface markers.