Single cell RNA sequencing (scRNA-seq) has become an established and powerful method to investigate transcriptomic cell-to-cell variation, revealing new cell types, and providing insights into developmental processes and transcriptional stochasticity. The array of published scRNA-seq protocols allow one to sequence transcriptomes from minute amounts of starting material. A key question is how these various protocols compare in terms of sensitivity of detection of mRNA molecules, and accuracy of quantification of expression. Here, we present an assessment of sensitivity and accuracy of many published data sets by spike-in standards with uniform data processing, including development of a flexible Unique Molecular Identifier (UMI) counting tool (https://github.com/vals/umis). We computationally compare 15 protocols, and experimentally assess 4 protocols on batch-matched cell populations, as well as investigating the impact of spike-in molecule degradation on two types of spike-ins. Our analysis provides an integrated framework for comparing different scRNA-seq protocols.

Differentiation of naïve CD4 + T cells into functionally distinct T helper subsets is crucial for the orchestration of immune responses. Due to extensive heterogeneity and multiple overlapping transcriptional programs in differentiating T cell populations, this process has remained a * Correspondence to: st9@sanger.ac.uk, Ashraful.Haque@qimrberghofer.edu.au or stegle@ebi.ac.uk. # denotes equal contribution † denotes equal contribution Author contributions TL and KRJ performed the single-cell RNA-seq experiments. VS developed the GPfates model in collaboration with MZ, NDL, OS and SAT. DFR and WRH generated the PbTII mouse model. KRJ, RM, IS, MSFS, LGF, ASN, UL, FSFG, PTB and CRE performed the mouse experiments. TL, VS, KRJ, LHL and FOB analysed the data and interpreted the results MJTS performed the TCR clonality analysis. TL, KRJ, RM, OB, AH and SAT designed the experiments. OS, AH and SAT cosupervised the study. TL, VS, KRJ, OS, AH and SAT wrote the manuscript. All authors have read and approved the manuscript. Competing interestsThe authors declare no competing interests. Data and materials availabilityThe data presented in this paper is publically available in the ArrayExpress database with accession number E-MTAB-4388. Europe PMC Funders Group Europe PMC Funders Author ManuscriptsEurope PMC Funders Author Manuscripts challenge for systematic dissection in vivo. By using single-cell transcriptomics and computational analysis using a temporal mixtures of Gaussian processes model, termed GPfates, we reconstructed the developmental trajectories of Th1 and Tfh cells during blood-stage Plasmodium infection in mice. By tracking clonality using endogenous TCR sequences, we first demonstrated that Th1/Tfh bifurcation had occurred at both population and single-clone levels. Next, we identified genes whose expression was associated with Th1 or Tfh fates, and demonstrated a T-cell intrinsic role for Galectin-1 in supporting a Th1 differentiation. We also revealed the close molecular relationship between Th1 and IL-10-producing Tr1 cells in this infection. Th1 and Tfh fates emerged from a highly proliferative precursor that upregulated aerobic glycolysis and accelerated cell cycling as cytokine expression began. Dynamic gene expression of chemokine receptors around bifurcation predicted roles for cell-cell in driving Th1/Tfh fates. In particular, we found that precursor Th cells were coached towards a Th1 but not a Tfh fate by inflammatory monocytes. Thus, by integrating genomic and computational approaches, our study has provided two unique resources, a database www.PlasmoTH.org, which facilitates discovery of novel factors controlling Th1/Tfh fate commitment, and more generally, GPfates, a modelling framework for characterizing cell differentiation towards multiple fates.

About half of the human genome consists of highly repetitive elements, most of which are considered dispensable for human life. Here, we report that repetitive elements originating from endogenous retroviruses (ERVs) are systematically transcribed during human early embryogenesis in a stage-specific manner. Our analysis highlights that the long terminal repeats (LTRs) of ERVs provide the template for stage-specific transcription initiation, thereby generating hundreds of co-expressed, ERV-derived RNAs. Conversion of human embryonic stem cells (hESCs) to an epiblast-like state activates blastocyst-specific ERV elements, indicating that their activity dynamically reacts to changes in regulatory networks. In addition to initiating stage-specific transcription, many ERV families contain preserved splice sites that join the ERV segment with non-ERV exons in their genomic vicinity. In summary, we find that ERV expression is a hallmark of cellular identity and cell potency that characterizes the cell populations in early human embryos.

Highthroughput single cell RNA sequencing (scRNAseq) has become an established and powerful method to investigate transcriptomic celltocell variation, and has revealed new cell types, and new insights into developmental process and stochasticity in gene expression. There are now several published scRNAseq protocols, which all sequence transcriptomes from a minute amount of starting material. Therefore, a key question is how these methods compare in terms of sensitivity of detection of mRNA molecules, and accuracy of quantification of gene expression. Here, we assessed the sensitivity and accuracy of many published data sets based on standardized spikeins with a uniform raw data processing pipeline. We developed a flexible and fast UMI counting tool (https://github.com/vals/umis) which is compatible with all UMI based protocols. This allowed us to relate these parameters to sequencing depth, and discuss the trade offs between the different methods. To confirm our results, we performed experiments on cells from the same population using three different protocols. We also investigated the effect of RNA degradation on spikein molecules, and the average efficiency of scRNAseq on spikein molecules versus endogenous RNAs.

Virus replication displays a large cell-to-cell heterogeneity; yet, not all sources of this variability are known. Here, we study the effect of defective interfering (DI) particle (DIP) co-infection on cell-to-cell variability in influenza A virus (IAV) replication. DIPs contain a large internal deletion in one of their eight viral RNAs (vRNA) and are, thus, defective in virus replication. Moreover, they interfere with virus replication. Using single-cell isolation and reverse transcription polymerase chain reaction, we uncovered a large between-cell heterogeneity in the DI vRNA content of infected cells, which was confirmed for DI mRNAs by single-cell RNA sequencing. A high load of intracellular DI vRNAs and DI mRNAs was found in low-productive cells, indicating their contribution to the large cell-to-cell variability in virus release. Furthermore, we show that the magnitude of host cell mRNA expression (some factors may inhibit virus replication), but not the ribosome content, may further affect the strength of single-cell virus replication. Finally, we show that the load of viral mRNAs (facilitating viral protein production) and the DI mRNA content are, independently from one another, connected with single-cell virus production. Together, these insights advance single-cell virology research toward the elucidation of the complex multi-parametric origin of the large cell-to-cell heterogeneity in virus infections.

In briefData analysis for single-cell transcriptomics requires sophisticated software pipelines. However, individual components of such a pipeline may be interdependent, and this interdependence is frequently overlooked. In this study, we investigate the interplay between an early step in single-cell transcriptome analysis, the imputation of missing data, and the later task of gene network reconstruction. We demonstrate that the choice of imputation method strongly influences the possible predicted network structures.