A Joint Model of RNA Expression and Surface Protein Abundance in Single Cells

Gayoso, Adam; Lopez, Romain; Steier, Zoë; Regier, Jeffrey; Streets, Aaron; Yosef, Nir

doi:10.1101/791947

Cited by 9 publications

(13 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In practice, users can choose more complex inference models if it is warranted for certain annotation types. Moreover, we expect prediction results to improve with more accurate and reproducible annotation methods, such as consistent cell type taxonomies provided by the Cell Ontology 48 project and better modeling of multimodal expression data 17 .…”

Section: Discussionmentioning

confidence: 99%

Efficient and precise single-cell reference atlas mapping with Symphony

Kang

Nathan

Zhang

et al. 2020

Preprint

View full text Add to dashboard Cite

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.

show abstract

Section: Discussionmentioning

confidence: 99%

Efficient and precise single-cell reference atlas mapping with Symphony

Kang

Nathan

Zhang

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…We next benchmarked WNN analysis against two recently introduced methods for multimodal integration: Multi-omics factor analysis v2 (MOFA+) [25], which uses a statistical framework based on factor analysis, and totalVI [26], which combines deep neural networks with a hierarchical Bayesian model. Both methods integrate the modalities into a latent space, which we used to construct an integrated k-NN graph and a 2D UMAP visualization.…”

Section: Wnn Analysis Is a Robust And Flexible Approach For Multimodamentioning

confidence: 99%

Integrated analysis of multimodal single-cell data

Hao

Andersen‐Nissen

et al. 2020

Preprint

375

474

View full text Add to dashboard Cite

The simultaneous measurement of multiple modalities, known as multimodal analysis, represents an exciting frontier for single-cell genomics and necessitates new computational methods that can define cellular states based on multiple data types. Here, we introduce "weighted-nearest neighbor analysis", an unsupervised framework to learn the relative utility of each data type in each cell, enabling an integrative analysis of multiple modalities. We apply our procedure to a CITE-seq dataset of hundreds of thousands of human white blood cells alongside a panel of 228 antibodies to construct a multimodal reference atlas of the circulating immune system. We demonstrate that integrative analysis substantially improves our ability to resolve cell states and validate the presence of previously unreported lymphoid subpopulations. Moreover, we demonstrate how to leverage this reference to rapidly map new datasets, and to interpret immune responses to vaccination and COVID-19. Our approach represents a broadly applicable strategy to analyze single-cell multimodal datasets, including paired measurements of RNA and chromatin state, and to look beyond the transcriptome towards a unified and multimodal definition of cellular identity. Availability: Installation instructions, documentation, tutorials, and CITE-seq datasets are available at http://www.satijalab.org/seurat

show abstract

“…These mixture models were fitted to the counts for each protein, while we used empty droplets to account for protein-specific background and mixture models to fit counts from all proteins within each droplet/cell to infer the technical component reflective of library size (Figs 1B-C). Another recent method defined protein expression as a mixture of biological cell statedependent foreground and noise-associated background 10 . It uses variational inference to learn the parameters of a probabilistic model that incorporates both latent cell-state variables and noise/technical factors.…”

Section: Discussionmentioning

confidence: 99%

Normalizing and denoising protein expression data from droplet-based single cell profiling

Mulè

Martins

Tsang

2020

Preprint

View full text Add to dashboard Cite

Recent methods enable simultaneous measurement of protein expression with the transcriptome in single cells by combining protein labeling with DNA barcoded antibodies followed by droplet based single cell capture and sequencing (e.g. CITE-seq). While data normalization and denoising have received considerable attention for single cell RNA-seq data, such methods for protein data have been less explored. Here we showed that a major source of noise in CITE-seq data originated from unbound antibody encapsulated in droplets. We also found that the counts of isotype controls and those of the "negative" population inferred from all protein counts of each cell are significantly correlated, suggesting that their covariation likely reflects cell-to-cell differences due to technical factors such as non-specific antibody binding and droplet-to-droplet differences in capture efficiency of the DNA tags. Motivated by these observations, we developed a normalization method for CITE-seq protein expression data called Denoised and Scaled by Background (DSB). DSB corrects for 1) protein-specific background noise as reflected by empty droplets, 2) the technical cell-to-cell variation as captured by the latent noise component described above. DSB normalization improves separation between positive and negative populations for each protein, centers the negative-staining population around zero, and can improve unbiased protein expression-based clustering. DSB is available through the open source R package "DSB" via a single function call and can be readily integrated with existing single cell analysis workflows, including those in Bioconductor and Seurat.

show abstract

A Joint Model of RNA Expression and Surface Protein Abundance in Single Cells

Cited by 9 publications

References 21 publications

Efficient and precise single-cell reference atlas mapping with Symphony

Efficient and precise single-cell reference atlas mapping with Symphony

Integrated analysis of multimodal single-cell data

Normalizing and denoising protein expression data from droplet-based single cell profiling

Contact Info

Product

Resources

About