Comparative analysis of single-cell RNA sequencing methods

Abstract

Single-cell RNA sequencing (scRNA-seq) offers exciting possibilities to address biological and medical questions, but a systematic comparison of recently developed protocols is still lacking. Here, we generated data from 447 mouse embryonic stem cells using Drop-seq, SCRB-seq, Smart-seq (on Fluidigm C1) and Smart-seq2 and analyzed existing data from 35 mouse embryonic stem cells prepared with CEL-seq. We find that Smart-seq2 is the most sensitive method as it detects the most genes per cell and across cells with the most even coverage, well suited for annotating transcriptomes. However, we also find that unique molecular identifiers (UMIs), available for CEL-seq, Drop-seq and SCRB-seq, reduce the measurement noise considerably, which is most relevant for quantifying transcriptomes. Importantly, we show by power simulations that SCRB-seq and Drop-seq are the most cost-efficient methods for detecting differentially expressed genes. Our analyses offer a solid basis for an informed choice among five prominent scRNA-seq protocols and for future evaluations of protocol improvements.