Modular and efficient pre-processing of single-cell RNA-seq

Melsted, Páll; Booeshaghi, A. Sina; Gao, Fan; Beltrame, Eduardo da Veiga; Moses, Lambda; Hjorleifsson, Kristjan E; Gehring, Jase; Pachter, Lior

doi:10.1101/673285

Cited by 102 publications

(97 citation statements)

References 46 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Alignment and UMI counting tools can affect the resulting expression matrix for gene expression modalities in scRNA-seq [9]. Furthermore, different tools have marked differences in their runtime and computation requirements [9,10]. To determine if this is also the case for UMI counting methods for ADT, we compared the runtime and assignment of sequencing reads to a unique CITE-seq antibody, cell barcode and UMI combination using three methods: CITE-seq-Count, Cell Ranger (running in featureOnly mode; 10X Genomics) and Kallisto-bustools (using the KITE pipeline) [10,11].…”

Section: Adt Counting Methods Have Minor Influence On the Resulting Cmentioning

confidence: 99%

“…Furthermore, different tools have marked differences in their runtime and computation requirements [9,10]. To determine if this is also the case for UMI counting methods for ADT, we compared the runtime and assignment of sequencing reads to a unique CITE-seq antibody, cell barcode and UMI combination using three methods: CITE-seq-Count, Cell Ranger (running in featureOnly mode; 10X Genomics) and Kallisto-bustools (using the KITE pipeline) [10,11]. To make the results more broadly applicable, in addition to the 52 antibody ADT library (ADT), we also included the 6 antibody cell hashing library used to demultiplex the samples (HTO) as well as three publicly available 17 antibody panel datasets where raw data is available (from the 10X Genomics website).…”

Section: Adt Counting Methods Have Minor Influence On the Resulting Cmentioning

confidence: 99%

See 1 more Smart Citation

Improving oligo-conjugated antibody signal in multimodal single-cell analysis

Buus

Herrera

Ivanova

et al. 2020

Preprint

View full text Add to dashboard Cite

Simultaneous measurement of surface proteins and gene expression within single cells offers highresolution snapshots of complex cell populations. These methods rely on staining cells with oligoconjugated antibodies analogous to staining for flow-and mass cytometry. Unlike flow-and mass cytometry, signal from oligo-conjugated antibodies is not hampered by spectral overlap or limited by the number of metal isotopes, making it a highly sensitive and scalable approach. Signal from oligoconjugated antibodies is quantified by counting reads from high-throughput sequencing. Consequently, cost of sequencing is strictly dependent on the signal intensities and background from the pool of antibodies used in analysis. Thus, considering the "cost-of-signal" as well as optimizing "signal-tonoise", makes titration of oligo-conjugated antibody panels more complex and even more important than for flow-and mass cytometry. In this study, we investigated the titration response of a panel of oligo-conjugated antibodies towards four variables: Antibody concentration, staining volume, cell number at staining, and tissue of origin. We find that staining with high antibody concentrations recommended by published protocols and commercial vendors cause unnecessarily high background signal and that concentrations of many antibodies can be drastically reduced without loss of biological information. Reducing staining volume only affects antibodies targeting highly abundant epitopes used at low concentrations and can be counteracted by reducing cell numbers at staining. We find that background signal from empty droplets can account for a major fraction of the total sequencing reads and is primarily derived from antibodies used at high concentrations. Together, this study provides new insight into the titration response and background signal of oligo-conjugated antibodies and offers concrete guidelines on how such panels can be improved.

show abstract

Section: Adt Counting Methods Have Minor Influence On the Resulting Cmentioning

confidence: 99%

Section: Adt Counting Methods Have Minor Influence On the Resulting Cmentioning

confidence: 99%

Improving oligo-conjugated antibody signal in multimodal single-cell analysis

Buus

Herrera

Ivanova

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…The SMART-Seq data was processed using kallisto with the `kallisto pseudo` command 23 . The 10x Genomics v3 data was pre-processed with kallisto and bustools 38 . Gene count matrices were made by using the --genecounts flag and TCC matrices were made by omitting it.…”

Section: Pre-processing Single-cell Rna-seq Datamentioning

confidence: 99%

Isoform cell type specificity in the mouse primary motor cortex

Booeshaghi

Yao

Velthoven

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Full-length SMART-Seq single-cell RNA-seq can be used to measure gene expression at isoform resolution, making possible the identification of isoform markers for cell types and for an isoform atlas. In a comprehensive analysis of 6,160 mouse primary motor cortex cells assayed with SMART-Seq, we find numerous examples of isoform specificity in cell types, including isoform shifts between cell types that are masked in gene-level analysis. These findings can be used to refine spatial gene expression information to isoform resolution. Our results highlight the utility of full-length single-cell RNA-seq when used in conjunction with other single-cell RNA-seq technologies.

show abstract

“…The kallisto bustools workflow [15][16][17] was used to obtain UMI [18] gene count matrices at different sampled read depths to mimic datasets sequenced at varying depths. From these subsamples, subsets of cells were sampled to emulate the sequencing of fewer cells (Figure 1).…”

Section: Resultsmentioning

confidence: 99%

Quantifying the tradeoff between sequencing depth and cell number in single-cell RNA-seq

Svensson

Beltrame

Pachter

2019

Preprint

Self Cite

View full text Add to dashboard Cite

The allocation of a sequencing budget when designing single cell RNA-seq experiments requires consideration of the tradeoff between number of cells sequenced and the read depth per cell. One approach to the problem is to perform a power analysis for a univariate objective such as differential expression. However, many of the goals of single-cell analysis requires consideration of the multivariate structure of gene expression, such as clustering. We introduce an approach to quantifying the impact of sequencing depth and cell number on the estimation of a multivariate generative model for gene expression that is based on error analysis in the framework of a variational autoencoder. We find that at shallow depths, the marginal benefit of deeper sequencing per cell significantly outweighs the benefit of increased cell numbers. Above about 15,000 reads per cell the benefit of increased sequencing depth is minor. Code for the workflow reproducing the results of the paper is available at https://github.com/pachterlab/SBP_2019/.

show abstract

Modular and efficient pre-processing of single-cell RNA-seq

Cited by 102 publications

References 46 publications

Improving oligo-conjugated antibody signal in multimodal single-cell analysis

Improving oligo-conjugated antibody signal in multimodal single-cell analysis

Isoform cell type specificity in the mouse primary motor cortex

Quantifying the tradeoff between sequencing depth and cell number in single-cell RNA-seq

Contact Info

Product

Resources

About