11

tSNE often offers better visual representation (separation) on such complicated data than PCA. As Micheal pointed out, computing a tSNE embedding over 20.000 gene dimensions is computationally unfeasible, so a number of PCs are normally calculated and these are used as input for calculating the tSNE. They are used in tandem. As for global vs. local, we are ...


10

Expected rates of doublets / duplets / multiplets Fluidigm C1 doublet rate: around 1-5% depending on chip type used. More information: Fluidigm white paper: Redesign of C1 Medium-Cell 96 IFCs Improves Single-Cell Capture Efficiency (PN 101-3328 B1) 10x Genomics Chromium 3'mRNA-Seq doublet rate: around 1% per 1000 cells captured Rates in the 'Chromium ...


8

It may be necessary to distinguish between methods that use unique molecular identifiers (UMIs), such as 10X's Chromium, Drop-seq, etc, and non-UMI methods, such as SMRT-seq. At least for UMI-based methods, the alternative perspective, that there is no significant zero-inflation in scRNA-seq, is also advocated in the single-cell research community. The ...


8

You only have 4 samples total. I think it would be difficult to not have the PCA show big differences between the groups with so few points. On the other hand, for differential expression, it is hard to get something to be statistically significant with only 2 replicates.


7

"Doublet" is commonly used to describe a droplet in droplet-based sequencing that has captured atleast 2 cells. 10x states their doublet rate to be 0.8% per 1000 cells: There is a tradeoff between targeted cell capture and doublet rate and lab protocols are usually optimized for a certain expected doublet rate. One should expect to see close to doubling ...


7

Is doublet a set of cells sequenced as a single cell? Yes. Depending on the method of single cell sequencing it may be more or less likely for groups of cells to be captured and barcoded with the same "unique" barcode. This is more likely in split-pool RNA sequencing (e.g. SPLiT-seq), and less likely in cell-capture RNA sequencing (e.g. Fluidigm ...


7

Single-cell analysis to compare samples is a long a difficult process. There is very good documentation for 10x Genomics cellranger, the DropSeq Pipeline and the Seurat R package. These tools all have GitHub repositories and the authors are very responsive if you encounter issues. Depending on the technology used to generate the data, you'll need to use ...


6

In the paper mentioned, we used the ScaleData function in Seurat to regress out the number of reads, Rn45s abundance, and percent ribosomal gene transcripts. Ribosomal genes were found with the regular expression ^Rp[sl][[:digit:]]. tiss <- ScaleData(object = tiss, vars.to.regress = c("nReads", "percent.ribo","Rn45s")) Here's a fuller notebook, and we'...


6

A SC3, single-cell consensus clustering, approach could be helpful here. It aims at achieving "high accuracy and robustness by combining multiple clustering solutions through a consensus approach" https://www.nature.com/nmeth/journal/v14/n5/full/nmeth.4236.html


6

The Biostars thread turned out helpful. The most interesting possible cause, not mentioned in the Ian Subery's answer, is that due to bursty nature of transcription, the true distribution of transcript counts across cells can be bimodal with a peak at zero even assuming a simple model of transcription such as the random telegraph model. See for example ...


6

According to Ilicic et al. (2016), on upregulation of mtRNA in broken cells: There is an extensive literature on the relationship between mtDNA, mitochondrially localized proteins, and cell death [34, 35]. However, upregulation of RNA levels of mtDNA in broken cells suggests losses in cytoplasmic content. In a situation where cell membrane is broken, ...


6

I do not know of such software. However, I believe this effort is a bit misdirected. The purpose of single-cell sequencing is to get a better understanding of cells; their heterogeneity and functional diversity or developmental / biological processes such as differentiation, using a higher "resolution" method. In other words, if we had the methods you are ...


6

I assume you are referring to VlnPlot() of Seurat. The reason that you are getting such a plot is because your distribution is highly skewed, most of your points are zeros, not surprising for scRNA-seq. And the plot is actually showing your distributions, it is just the points at zero stacking on top of each other. You can double check this by plotting a ...


5

I don't think you can conclude that the dataset is terrible based on that PCA. Depending on the specific protocol, each scRNA-seq dataset is going to be very different. Unlike bulk RNA-seq where all the samples are going to be of very similar quality, individual cells will be highly variable. For example, the most basic QC metric is the number of reads. If ...


5

I know of no references for this, but in general, I would say that your reasoning is sound. I would just add that in contrast to what I suspect you have simulated, not all transcripts are equally likely to be captured and amplified. We don't really understand what the determinants of this are, but for example, GC content is definitely related.


5

Look at this recent paper that uses ComBat on scRNA-seq data for batch effect removal and states that it "successfully does so". I also suggest that you check out this publication on Distribution Matching Residual-Nets. Authors evaluated their method also on scRNA-seq data and thus it may be something you are looking for. I personally played a bit with ...


5

We've found ribosomal RNA to be less of a problem with sequencing that depends on polyA, which suggests the issue might be in the library preparation, rather than the selection. Many polyA RNA library preparation methods involve amplification, rather than selection, which means that existing transcripts that are present in very high abundance (such as rRNA) ...


5

From the page you cited: During first-strand synthesis, upon reaching the 5’ end of the RNA template, the terminal transferase activity of the MMLV reverse transcriptase adds a few additional nucleotide That usually are CCC (1st is almost always C and the other two most of the times), see: Zajac et al., 2013 These CCCs are used for priming the Template ...


5

I'm unsure whether this is the answer you are looking for, but when looking into 10X cellranger documentation for the Matrices Output: Unfiltered gene-barcode matrices: Contains every barcode from fixed list of known-good barcode sequences. This includes background and non-cellular barcodes. Filtered gene-barcode matrices: Contains only detected ...


5

Seruat will give you a list of genes which it thinks are upregulated in a particular cluster. Look at the functions that talk about marker genes - these functions basically do a DE analysis of the genes in one cluster compared to the others. Then take that list and feed it to any standard GO analysis tool. Have a look at the topGO topKEGG and geneSetTest ...


5

I don't think this is possible in Seurat v2, but in v3 you can change the factor levels of the grouping variable to change the plot order: library(Seurat) FeaturePlot(object = pbmc_small, features = head(VariableFeatures(pbmc_small), 2), split.by = 'groups') Change the order: pbmc_small$groups <- factor(pbmc_small$groups, ...


5

This can be solved like this: library(Seurat) my_genes <- c("gene1", "gene2", "gene3") exp <- FetchData(object, my_genes) matrix <- as.matrix(colMeans(exp > 0))*100


5

Basically what you're discovering is that there are unannotated expressed features, so your task isn't really finding peaks, but rather finding novel expressed transcripts. For that, you can use stringTie or similar programs. Ensure that the BAM file you give to stringTie have all of the cells together.


5

These tutorials on Seurat multimodal data and the wrapper Seurat data are easy ways to start. The wrapper has some cite-seq data preinstalled making it easy to work with benchmarked data sets If you are using Seurat u can just as well have a look at Signac to start working with some sc-atac seq


5

The simplest would be using a count matrix (at the end of the link you have shared, section "Supplementary file"). For example GSE138651_Vagal_WholeNodoseSeq_raw.mtx.gz corresponds to "raw" count matrices obtained after mapping and counting the reads. This file, along with the GSE138651_barcodes.tsv.gz and GSE138651_genes.tsv.gz can be used to "read" the ...


5

Ah, looks like I can't even procrastinate on StackExchange anymore without seeing work-related stuff. Oh well. Anyway, the other answers and comments are way off. scran has supported sparse matrices for years, ever since we switched over to the SingleCellExperiment class as our basic data structure. quickCluster does no coercion to dense format unless you ...


5

For merging dataframes, I find it easiest to use the tidyverse / dplyr functions inner/full/left/right_join. See the "Data Transformation Cheatsheet" on this page. For the merge that @user438383 has mentioned, this would be a left_join: library(tidyverse) anno <- data.frame(cluster = 1:6, celltype = c("T:CD4+NAIVE", "T:CD4+NAIVE&...


4

Data preparation Cell Ranger uses the Illumina sequencing output (.bcl) files Make fastq files: cellranger mkfastq ==> .fastq Prepare count matrix: cellranger count ==> matrix.mtx, web_summary.html, cloupe.cloupe Optional: combine multiple matrix.mtx files (libraries): cellranger aggr Data analysis Loupe Cell Browser visualization of cloupe.cloupe files ...


4

My primary concern with using the top ~1% or so in upper quantile normalization is that it's going to be prone to the same robustness issues that RPKM/FPKMs run in to. That is, if for whatever technical reason you have to have a fair bit of variability in a couple very highly expressed genes (typically rRNAs, but one can imagine other genes) and the set of ...


4

We cannot assume that doublets will produce more UMIs I would caution the assumption that all doublets will have twice the UMI levels of isolated single-cells. Many "doublets" could contain multiplets of 3 or more cells, depending on how many cells have been loaded into the experiment. Most vendors of single-cell technologies quote multiplet rates for ...


Only top voted, non community-wiki answers of a minimum length are eligible