Semisoft clustering of single-cell data

Zhu, Lingxue; Lei, Jing; Klei, Lambertus; Devlin, Bernie; Roeder, Kathryn

doi:10.1073/pnas.1817715116

Cited by 77 publications

(69 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“… a , The optimal number of mesenchymal cell clusters was determined using the SOUP method 15 , a semi-soft clustering algorithm designed to distinguish between distinct cell types and transition states between cell types. b , Main cluster identity from SOUP highlighted on the t-SNE from figure 1b (mesenchymal cell types only).…”

Section: Figure S1mentioning

confidence: 99%

Combined single-cell and spatial transcriptomics reveals the molecular, cellular and spatial bone marrow niche organization

Baccin

Al-Sabah

Velten

et al. 2019

Preprint

View full text Add to dashboard Cite

36The bone marrow (BM) constitutes the primary site for life-long blood production and skeletal 37 regeneration. However, its cellular composition and the spatial organization into distinct 38 'niches' remains controversial. Here, we combine single-cell and spatially resolved 39 transcriptomics to systematically map the molecular and cellular composition of the endosteal, 40 sinusoidal, and arteriolar BM niches. This allowed us to transcriptionally profile all major BM 41 resident cell types, determine their localization, and clarify the cellular and spatial sources of 42 key growth factors and cytokines. Our data demonstrate that previously unrecognized Cxcl12-43 abundant reticular (CAR) cell subsets (i.e. Adipo-and Osteo-CAR cells) differentially localize 44 to sinusoidal or arteriolar surfaces, locally act as 'professional cytokine secreting cells', and 45 thereby establish distinct peri-vascular micro-niches. Importantly, we also demonstrate that the 46 3-dimensional organization of the BM can be accurately inferred from single-cell gene 47 expression data using the newly developed RNA-Magnet algorithm. Together, our study reveals 48 the cellular and spatial organization of BM niches, and offers a novel strategy to dissect the 49 complex organization of whole organs in a systematic manner. 50 51 One Sentence Summary: Integration of single-cell and spatial transcriptomics reveals the 52 molecular, cellular and spatial organization of bone marrow niches 53 54 65 endosteal niches 4-10 . To gain a global understanding of cell types and niches in the BM, we 66 have generated a single-cell RNA-sequencing (scRNAseq)-based molecular map of all major 67 BM populations. We then used spatially resolved transcriptomics in combination with novel 68 3 computational tools to allocate cell types to different BM niches, determine molecular 69 mediators of intercellular interactions, and identify the cellular and spatial sources of niche 70 factors. 71 72 RESULTS 73 Identification and characterization of BM resident cell types by scRNAseq 74 Frequencies of BM cell types differ by several orders of magnitude 11 , imposing a challenge to 75 scRNAseq approaches. To capture both highly abundant as well as extremely rare BM resident 76 cells, we performed droplet-based scRNAseq 12 of cells from total mouse BM, followed by 77 progressive depletion of highly abundant cell types or enrichment of rare populations from 78 undigested BM or enzymatically digested bones (Figure 1a). In total, our dataset comprises 79 7497 cells with a median detection of 1999 genes per single cell, which formed 32 clusters 80 corresponding to distinct cell types or stages of differentiation (Figure 1b, S1). Importantly, this 81 map is not quantitative with regard to the relative size of the different cell populations, since 82 dissociation rates largely differ between cell types 11 . As detailed below, cell type annotation 83 was performed based on marker gene expression (Table S1, Figure S2,3), gene ontology 84 analyses (Table S1), and by quantifyin...

show abstract

Section: Figure S1mentioning

confidence: 99%

Combined single-cell and spatial transcriptomics reveals the molecular, cellular and spatial bone marrow niche organization

Baccin

Al-Sabah

Velten

et al. 2019

Preprint

View full text Add to dashboard Cite

show abstract

“…Besides PCA, other DR techniques are also commonly used for cell clustering. For example, nonnegative matrix factorization (NMF) is used in SOUP [19]. Partial least squares is used in scPLS [20].…”

Section: Introductionmentioning

confidence: 99%

Accuracy, Robustness and Scalability of Dimensionality Reduction Methods for Single Cell RNAseq Analysis

Sun

Zhu

et al. 2019

Preprint

View full text Add to dashboard Cite

Background: Dimensionality reduction (DR) is an indispensable analytic component for many areas of single cell RNA sequencing (scRNAseq) data analysis. Proper DR can allow for effective noise removal and facilitate many downstream analyses that include cell clustering and lineage reconstruction. Unfortunately, despite the critical importance of DR in scRNAseq analysis and the vast number of DR methods developed for scRNAseq studies, however, few comprehensive comparison studies have been performed to evaluate the effectiveness of different DR methods in scRNAseq.Results: Here, we aim to fill this critical knowledge gap by providing a comparative evaluation of a variety of commonly used DR methods for scRNAseq studies. Specifically, we compared 18 different DR methods on 30 publicly available scRNAseq data sets that cover a range of sequencing techniques and sample sizes. We evaluated the performance of different DR methods for neighborhood preserving in terms of their ability to recover features of the original expression matrix, and for cell clustering and lineage reconstruction in terms of their accuracy and robustness. We also evaluated the computational scalability of different DR methods by recording their computational cost. Conclusions:Based on the comprehensive evaluation results, we provide important guidelines for choosing DR methods for scRNAseq data analysis. We also provide all analysis scripts used in the present study at www.xzlab.org/reproduce.html. Together, we hope that our results will serve as an important practical reference for practitioners to choose DR methods in the field of scRNAseq analysis. our results can serve as an important guideline for practitioners to choose DR methods in the field of scRNAseq analysis.

show abstract

“…We recommend that users consider performing clustering experiments on 500 to 1000 highly variable genes. In addition to CIDR and SIMLR, we also compared scDMFK with other commonly used scRNA-seq data clustering methods, such as Seurat (Satija et al, 2015), SC3 (Kiselev et al, 2017), Raceid (Herman and Grün, 2018), and SOUP (Zhu et al, 2019). Considering that the Seurat method cannot give a specific number of clusters in advance, we ran it several times with its parameter "resolution" changing from 0.5 to 1.5 by 0.1 and took the best ARI and NMI value as its result and recorded the corresponding estimated cluster number.…”

Section: Discussionmentioning

confidence: 99%

“…Most existing clustering algorithms customized for single-cell analysis do not model and denoise such data. Typically, they first learn the predefined distance measure and similarity metric based on an original data matrix directly or a reduced data matrix by simple linear dimension reduction methods, like PCA and ICA, and then utilize the traditional hard clustering methods, such as standard k-means clustering (Marco et al, 2014;Grün et al, 2016), graph-based spectral clustering (Wang et al, 2017;Zhu et al, 2019) or community detection (Levine et al, 2015;Satija et al, 2015), density-based clustering (Jiang et al, 2016), integrated learning clustering (Kiselev et al, 2017;Yang et al, 2018), and hierarchical clustering (Zeisel et al, 2015;Lin et al, 2017). However, in addition to the possibility of spurious identification of cell subtypes by separating dimension reduction and clustering, expensive computation limits their performance on large-scale datasets.…”

Section: Introductionmentioning

confidence: 99%

Single-Cell Transcriptome Data Clustering via Multinomial Modeling and Adaptive Fuzzy K-Means Algorithm

et al. 2020

View full text Add to dashboard Cite

Semisoft clustering of single-cell data

Cited by 77 publications

References 32 publications

Combined single-cell and spatial transcriptomics reveals the molecular, cellular and spatial bone marrow niche organization

Combined single-cell and spatial transcriptomics reveals the molecular, cellular and spatial bone marrow niche organization

Accuracy, Robustness and Scalability of Dimensionality Reduction Methods for Single Cell RNAseq Analysis

Single-Cell Transcriptome Data Clustering via Multinomial Modeling and Adaptive Fuzzy K-Means Algorithm

Contact Info

Product

Resources

About