As the number of single‐cell transcriptomics datasets grows, the natural next step is to integrate the accumulating data to achieve a common ontology of cell types and states. However, it is not straightforward to compare gene expression levels across datasets and to automatically assign cell type labels in a new dataset based on existing annotations. In this manuscript, we demonstrate that our previously developed method, scVI, provides an effective and fully probabilistic approach for joint representation and analysis of scRNA‐seq data, while accounting for uncertainty caused by biological and measurement noise. We also introduce single‐cell ANnotation using Variational Inference (scANVI), a semi‐supervised variant of scVI designed to leverage existing cell state annotations. We demonstrate that scVI and scANVI compare favorably to state‐of‐the‐art methods for data integration and cell state annotation in terms of accuracy, scalability, and adaptability to challenging settings. In contrast to existing methods, scVI and scANVI integrate multiple datasets with a single generative model that can be directly used for downstream tasks, such as differential expression. Both methods are easily accessible through scvi‐tools.

Selection, recombination, and hybrid evolution: Hybridization is an important force in evolution. The effects of hybridization across the whole genome are not understood. Using a fine-scale genetic map, Schumer et al. examined local ancestry in replicate natural hybrid populations of swordtail fish. Each parental species contributes different proportions of genetic material to the genomes of their descendants. Genes from the “minor” (less well-represented) parent occur in regions of the genome that are subject to higher recombination rates and where there are fewer potentially deleterious genes. Neanderthal ancestry in human genomes shows similar patterns. Science , this issue p. 656

Cerebrospinal fluid (CSF) protects the central nervous system (CNS) and analyzing CSF aids the diagnosis of CNS diseases, but our understanding of CSF leukocytes remains superficial. Here, using single cell transcriptomics, we identify a specific location-associated composition and transcriptome of CSF leukocytes. Multiple sclerosis (MS) – an autoimmune disease of the CNS – increases transcriptional diversity in blood, but increases cell type diversity in CSF including a higher abundance of cytotoxic phenotype T helper cells. An analytical approach, named cell set enrichment analysis (CSEA) identifies a cluster-independent increase of follicular (TFH) cells potentially driving the known expansion of B lineage cells in the CSF in MS. In mice, TFH cells accordingly promote B cell infiltration into the CNS and the severity of MS animal models. Immune mechanisms in MS are thus highly compartmentalized and indicate ongoing local T/B cell interaction.

The abundance of new computational methods for processing and interpreting transcriptomes at a single cell level raises the need for in silico platforms for evaluation and validation. Here, we present SymSim, a simulator that explicitly models the processes that give rise to data observed in single cell RNA-Seq experiments. The components of the SymSim pipeline pertain to the three primary sources of variation in single cell RNA-Seq data: noise intrinsic to the process of transcription, extrinsic variation indicative of different cell states (both discrete and continuous), and technical variation due to low sensitivity and measurement noise and bias. We demonstrate how SymSim can be used for benchmarking methods for clustering, differential expression and trajectory inference, and for examining the effects of various parameters on their performance. We also show how SymSim can be used to evaluate the number of cells required to detect a rare population under various scenarios.