Efficient and precise single-cell reference atlas mapping with Symphony

Abstract

Recent advances in single-cell technologies and integration algorithms make it possible to construct large, comprehensive reference atlases from multiple datasets encompassing many donors, studies, disease states, and sequencing platforms. Much like mapping sequencing reads to a reference genome, it is essential to be able to map new query cells onto complex, multimillion-cell reference atlases to rapidly identify relevant cell states and phenotypes. We present Symphony, a novel algorithm for building compressed, integrated reference atlases of ≥106 cells and enabling efficient query mapping within seconds. Based on a linear mixture model framework, Symphony precisely localizes query cells within a low-dimensional reference embedding without the need to reintegrate the reference cells, facilitating the downstream transfer of many types of reference-defined annotations to the query cells. We demonstrate the power of Symphony by (1) mapping a query containing multiple levels of experimental design to predict pancreatic cell types in human and mouse, (2) localizing query cells along a smooth developmental trajectory of human fetal liver hematopoiesis, and (3) harnessing a multimodal CITE-seq reference atlas to infer query surface protein expression in memory T cells. Symphony will enable the sharing of comprehensive integrated reference atlases in a convenient, portable format that powers fast, reproducible querying and downstream analyses.

Cross-tissue, single-cell stromal atlas identifies shared pathological fibroblast phenotypes in four chronic inflammatory diseases

“…We wanted to compare dermal fibroblasts directly to clusters defined in our fibroblast atlas. To do this, we leveraged a novel algorithm, Symphony (Kang et al, 2020) ( Methods ), designed to quickly and accurately map new scRNAseq profiles into a harmonized atlas to compare them with annotated reference cells. Using Symphony, we mapped dermal fibroblasts into our multi-tissue fibroblast atlas and projected them into the reference UMAP space for visual comparison ( Figure 6c ).…”

“…We used a novel type of analysis from single-cell analysis called Symphony reference mapping (Kang et al, 2020) to compare human dermal fibroblasts and mouse lung, synovial, and lung fibroblasts to our annotated cross-tissue atlas. Reference mapping let us avoid intensive and error-prone manual interpretation steps in de novo analysis of the external datasets.…”

“…K.W. is supported by a NIH-NIAMS Clinical Investigator Award (1K08AR077037-01) and a BWH The Symphony pipeline is described in detail in a separate manuscript (Kang et al, 2020). In order to infer reference cluster identity in query cells, we used a k-NN classifier.…”

“…The identification of shared cell states across tissues with scRNAseq has recently become possible with advances in statistical methods for integrative clustering (Butler et al, 2018; Korsunsky et al, 2019; Tran et al, 2020) and reference mapping (Andreatta et al, 2020; Kang et al, 2020; Lotfollahi et al, 2020). Integrative clustering identifies similar cell states across a range of scRNAseq datasets, even when the datasets come from different donors, species, or tissues.…”

Big data analytics in single‐cell transcriptomics: Five grand opportunities

“…There lies an opportunity for creating a central repository for integrating the different studies and storing the big data in a homogenous format. Databases such as PanglaoDB (Franzén, Gan, & Björkegren, 2019) and workflows such as recount (Collado‐Torres, Nellore, Kammers, & Ellis, 2016), Symphony (Kang et al, 2020), LIGER (Welch et al, 2019), Seurat (Butler, Hoffman, Smibert, Papalexi, & Satija, 2018), and Harmony integrate data generated from multiple studies and present them through a unified framework to make a computational method more effective.…”