Characterisation of CD4+\u2009T-cell subtypes using single cell RNA sequencing and the impact of cell number and sequencing depth

James Ding,Paul Martin,Gisela Orozco,Anne Barton,Samantha L Smith,Steve Eyre

doi:10.1038/s41598-020-76972-9

Abstract

CD4+ T-cells represent a heterogeneous collection of specialised sub-types and are a key cell type in the pathogenesis of many diseases due to their role in the adaptive immune system. By investigating CD4+ T-cells at the single cell level, using RNA sequencing (scRNA-seq), there is the potential to identify specific cell states driving disease or treatment response. However, the impact of sequencing depth and cell numbers, two important factors in scRNA-seq, has not been determined for a complex cell population such as CD4+ T-cells. We therefore generated a high depth, high cell number dataset to determine the effect of reduced sequencing depth and cell number on the ability to accurately identify CD4+ T-cell subtypes. Furthermore, we investigated T-cell signatures under resting and stimulated conditions to assess cluster specific effects of stimulation. We found that firstly, cell number has a much more profound effect than sequencing depth on the ability to classify cells; secondly, this effect is greater when cells are unstimulated and finally, resting and stimulated samples can be combined to leverage additional power whilst still allowing differences between samples to be observed. While based on one individual, these results could inform future scRNA-seq studies to ensure the most efficient experimental design.

Highlights

Whilst genetics studies have been successful in identifying single nucleotide polymorphisms (SNPs) associated with common complex disease susceptibility, mortality and o utcome[1,2,3], they have had limited impact for predicting treatment response and there is currently great interest in discovering biomarkers which can predict if a patient will respond to a given therapy[4]
First, it was possible to accurately define cell clusters in unstimulated cells at read depths of approximately 60–70 thousand reads per cell; second, stimulated cells showed a higher classification accuracy and cell number had a much more profound effect, requiring a minimum of approximately 2500 cells for accurate classification
The effect of sequencing read depth and cell numbers have previously been studied for single cell RNA-seq[16,17]

Summary

Introduction

Whilst genetics studies have been successful in identifying single nucleotide polymorphisms (SNPs) associated with common complex disease susceptibility, mortality and o utcome[1,2,3], they have had limited impact for predicting treatment response and there is currently great interest in discovering biomarkers which can predict if a patient will respond to a given therapy[4]. Single cell genomic approaches, such as mass cytometry by time of flight (CyTOF) and single cell RNA-seq (scRNA-seq), have the potential to fully explore this heterogeneity by independently assaying individual cells This can help disentangle the population substructure and identify differences, such as rare populations or changes in sub-type frequency, between two conditions. The development of droplet-based systems, such as Drop-Seq[11] or the 10x Genomics Chromium C ontroller[12], allows researchers to study thousands of cells, overcoming the limitation of cell number in lower throughput microfluidic or plate-based techniques This allows the accurate profiling of more complex cell populations in a high throughput, cost-effective manner

Objectives

Methods

Results

Conclusion