Systematic comparison of single-cell and single-nucleus RNA-sequencing methods

Ding, Jiarui; Adiconis, Xian; Simmons, Sean; Kowalczyk, Monika S.; Hession, Cynthia C.; Marjanovic, Nemanja D.; Hughes, Travis; Wadsworth, Marc H.; Burks, Tyler; Nguyễn, Lan; Kwon, John; Barak, Boaz; Ge, William; Kedaigle, Amanda J.; Carroll, Shaina L.; Li, Shuqiang; Hacohen, Nir; Rozenblatt-Rosen, Orit; Shalek, Alex K.; Villani, Alexandra‐Chloé; Regev, Aviv; Levin, Joshua Z.

doi:10.1038/s41587-020-0465-8

Cited by 673 publications

(732 citation statements)

References 48 publications

Supporting

Mentioning

653

Contrasting

Unclassified

Order By: Relevance

“…The megakaryocytes from the eQTL dataset are for instance seen as a subpopulation of the megakaryocytes of the Bench10Xv2 dataset. Different marker genes were used to identify these populations, so it could indeed be that one set of marker genes was more specific [24,25]. Looking at a UMAP embedding of the aligned datasets, we also notice that the two populations do not completely overlap ( Figure 6C-D).…”

Section: Linear Svm Can Learn the Classification Tree During An Intermentioning

confidence: 93%

See 1 more Smart Citation

Hierarchical progressive learning of cell identities in single-cell data

Michielsen

Reinders

Mahfouz

2020

Preprint

View full text Add to dashboard Cite

Motivation:In single-cell RNA-sequencing datasets, cell identification is mainly done manually, which is subjective and time-consuming. As a consequence, most datasets are annotated at a different resolution. This is not surprising as cell populations form a hierarchy, but it can be problematic for downstream analysis or comparison of datasets. Several supervised methods have been developed to overcome the drawbacks of unsupervised learning, but none of these combines the information of multiple datasets and preserves the old definition of the cell populations in each dataset. Results: To overcome these challenges, we developed a hierarchical progressive learning method which automatically finds relationships between populations of multiple datasets and uses this to construct a classification tree. We implemented the tree with a one-class and linear SVM for each node and evaluated the classification performance, including the rejection option, and tree construction. At the moment, choosing between a one-class and linear SVM is a trade-off between the ability of discovering new cell populations and a higher accuracy. Both the one-class and linear SVM also outperform other hierarchical classifiers. Furthermore, we show that it is possible to construct a correct classification tree for immune cells when combining three PBMC datasets and predict the labels of the fourth dataset with high accuracy.

show abstract

Section: Linear Svm Can Learn the Classification Tree During An Intermentioning

confidence: 93%

“…Each cell population consists of 2,000 cells. The total dataset consists of 20,000 cells and 21,952 genes.The PBMC-Bench10Xv2 and PBMC-Bench10Xv3 datasets are the PbmcBench pbmc1.10Xv2 and pbmc1.10Xv3 datasets from[25]. These datasets consist of 6,444 and 3,222 cells respectively, 22,280 genes, and nine different cell populations.…”

mentioning

confidence: 99%

Hierarchical progressive learning of cell identities in single-cell data

Michielsen

Reinders

Mahfouz

2020

Preprint

View full text Add to dashboard Cite

show abstract

“…The increased statistical power of BARseq, obtained at the cost of some spatial resolution, is reminiscent of different clustering power across single-cell RNAseq techniques of varying throughput and read depth (Ding et al, 2020;Yao et al, 2020). The high throughput of BARseq thereby provides a powerful asset for investigating the organization of projection patterns and their relationship to gene expression.…”

Section: Barseq2 Reveals Gene Correlates Of Projectionsmentioning

confidence: 99%

Integrating barcoded neuroanatomy with spatial transcriptional profiling reveals cadherin correlates of projections shared across the cortex

Sun

Chen

Fischer

et al. 2020

Preprint

View full text Add to dashboard Cite

Functional circuits consist of neurons with diverse axonal projections and gene expression. Understanding the molecular signature of projections requires high-throughput interrogation of both gene expression and projections to multiple targets in the same cells at cellular resolution, which is difficult to achieve using current technology. Here, we introduce BARseq2, a technique that simultaneously maps projections and detects multiplexed gene expression by in situ sequencing. We determined the expression of cadherins and cell-type markers in 29,933 cells, and the projections of 3,164 cells in both the mouse motor cortex and auditory cortex. Associating gene expression and projections in 1,349 neurons revealed shared cadherin signatures of homologous projections across the two cortical areas. These cadherins were enriched across multiple branches of the transcriptomic taxonomy. By correlating multi-gene expression and projections to many targets in single neurons with high throughput, BARseq2 provides a path to uncovering the molecular logic underlying neuronal circuits.

show abstract

“…Integrating inDrop data from ( 6 ) and 10X Chromium 5' data required addressing the systematic differences 51 in gene capture present between these technologies and 10X Chromium 3' data that was used to develop the clustering model. Analysis of the differences in gene expression between the technologies suggested that a multiplicative correction factor per each gene i could adjust for the capture efficiency differences.…”

Section: Integration Of Additional Single-cell Datamentioning

confidence: 99%

CITEseq analysis of non-small-cell lung cancer lesions reveals an axis of immune cell activation associated with tumor antigen load andTP53mutations

Leader

Grout

Chang

et al. 2020

Preprint

View full text Add to dashboard Cite

Immunotherapy is becoming a mainstay in the treatment of NSCLC. While tumor mutational burden (TMB) has been shown to correlate with response to immunotherapy, little is known about the relation of the baseline immune response with the tumor genotype. Here, we profiled 35 early stage NSCLC lesions using multiscale single cell sequencing. Unsupervised clustering identified in a subset of patients a key cellular module consisting of PDCD1+ CXCL13+ activated T cells, IgG+ plasma cells, and SPP1+ macrophages, referred to as the lung cancer activation module (LCAMhi). Transcriptional data from two NSCLC cohorts confirmed a subset of patients with LCAMhi enrichment, which was independent of overall immune cell content. The LCAMhi module strongly correlated with TMB, expression of cancer testis antigens, and with TP53 mutations in smokers and non-smokers. These data establish LCAM as a key mode of immune cell activation associated with high tumor antigen load and driver mutations.

show abstract

Systematic comparison of single-cell and single-nucleus RNA-sequencing methods

Cited by 673 publications

References 48 publications

Hierarchical progressive learning of cell identities in single-cell data

Hierarchical progressive learning of cell identities in single-cell data

Integrating barcoded neuroanatomy with spatial transcriptional profiling reveals cadherin correlates of projections shared across the cortex

CITEseq analysis of non-small-cell lung cancer lesions reveals an axis of immune cell activation associated with tumor antigen load andTP53mutations

Contact Info

Product

Resources

About