Evaluating the practical aspects and performance of commercial single-cell RNA sequencing technologies

The rapid development of updated and new commercially available single-cell transcriptomics platforms is providing a range of experimental options. Cost, sensitivity, throughput, flexibility, and ease of use influence that selection. We performed a comprehensive comparison of single-cell transcriptomic approaches using multiple standardized PBMCs. In evaluating seven recently available kits that interrogate whole transcriptome mRNA and two including TCR sequencing, we report on standard single-cell sensitivity metrics including gene saturation, UMI detection, cell recovery, sequencing efficiency, and cell subset resolution. We also discuss workflow throughput, sample requirements, timing, cost, and labor as critical factors to consider based on project goals. In addition to variable experimental constraints imposed by each platform, our findings highlight differences in cell recovery and sensitivity, which we found to significantly influence cell subset resolution. This work provides a basis by which users can balance performance and practical considerations when selecting an RNA single-cell sequencing platform. For our first experiment, Benchmark 1 (BM1), we used two healthy PBMC donor samples in replicate, to compare seven platforms: 10x Genomics’ Chromium NextGEM 3′ v3.1 (labeled “NextGEM 3’” in figures and results), GEM-X 3’ v4 (GEM-X 3’), and Fixed RNA Profiling (Flex), Parse’s Evercode WT v.3 (Parse v3), Scale Biosciences Single Cell RNA kit (Scale), and Fluent Biosciences PIP Seq T20 3′ v4 (Fluent v4) and PIPseq V T20 3’ (Fluent V). Each of these kits target 3′ and whole transcriptome RNA from single cells. All platforms were run in parallel on the same day except for Scale and Fluent PIPseq V assays, which were processed separately using aliquots from the same donors’ leukapheresis samples (Figure 1a). For our second experiment, Benchmark 2 (BM2), we compared platforms that generate TCR repertoire profiling (Supplemental Figure 1a). For this, we sorted T-cells based on CD3 expression from 4 different healthy donor PBMCs, with overlap of two BM1 donors, to run on 10x Genomic’s Chromium NextGEM 5′ v2 (NextGEM 5′), GEM-X 5′ v3 (GEM-X 5’) and Parse Evercode v.2 TCR (Parse v2).