Description Usage Arguments Details Value Author(s) References See Also Examples
Identify potential doublet cells based on simulations of putative doublet expression profiles.
This function is now deprecated, use computeDoubletDensity
from scDblFinder instead.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25  doubletCells(x, ...)
## S4 method for signature 'ANY'
doubletCells(
x,
size.factors.norm = NULL,
size.factors.content = NULL,
k = 50,
subset.row = NULL,
niters = max(10000, ncol(x)),
block = 10000,
d = 50,
force.match = FALSE,
force.k = 20,
force.ndist = 3,
BNPARAM = KmknnParam(),
BSPARAM = bsparam(),
BPPARAM = SerialParam()
)
## S4 method for signature 'SummarizedExperiment'
doubletCells(x, ..., assay.type = "counts")
## S4 method for signature 'SingleCellExperiment'
doubletCells(x, size.factors.norm = sizeFactors(x), ...)

x 
A numeric matrixlike object of count values, where each column corresponds to a cell and each row corresponds to an endogenous gene. Alternatively, a SummarizedExperiment or SingleCellExperiment object containing such a matrix. 
... 
For the generic, additional arguments to pass to specific methods. For the SummarizedExperiment and SingleCellExperiment methods, additional arguments to pass to the ANY method. 
size.factors.norm 
A numeric vector of size factors for normalization of For the SingleCellExperiment method, the default values are taken from 
size.factors.content 
A numeric vector of size factors for RNA content normalization of 
k 
An integer scalar specifying the number of nearest neighbours to use to determine the bandwidth for density calculations. 
subset.row 
See 
niters 
An integer scalar specifying how many simulated doublets should be generated. 
block 
An integer scalar controlling the rate of doublet generation, to keep memory usage low. 
d 
An integer scalar specifying the number of components to retain after the PCA. 
force.match 
A logical scalar indicating whether remapping of simulated doublets to original cells should be performed. 
force.k 
An integer scalar specifying the number of neighbours to use for remapping if 
force.ndist 
A numeric scalar specifying the bandwidth for remapping if 
BNPARAM 
A BiocNeighborParam object specifying the nearest neighbor algorithm.
This should be an algorithm supported by 
BSPARAM 
A BiocSingularParam object specifying the algorithm to use for PCA, if 
BPPARAM 
A BiocParallelParam object specifying whether the neighbour searches should be parallelized. 
assay.type 
A string specifying which assay values contain the count matrix. 
This function simulates doublets by adding the count vectors for two randomly chosen cells in x
.
For each original cell, we compute the density of neighboring simulated doublets and compare it to the density of neighboring original cells.
Genuine doublets should have a high density of simulated doublets relative to the density of its neighbourhood.
Thus, the doublet score for each cell is defined as the ratio of densities of simulated doublets to the (squared) density of the original cells.
Densities are calculated in lowdimensional space after a PCA on the lognormalized expression matrix of x
.
Simulated doublets are projected into the lowdimensional space using the rotation vectors computed from the original cells.
A tricube kernel is used to compute the density around each cell.
The bandwidth of the kernel is set to the median distance to the k
nearest neighbour across all cells.
The two size factor arguments have different roles:
size.factors.norm
contains the size factors to be used for normalization prior to PCA and distance calculations.
This defaults to the values returned by librarySizeFactors
but can be explicitly set to ensure that the lowdimensional space is consistent with that in the rest of the analysis.
size.factors.content
is much more important, and represents the size factors that preserve RNA content differences.
This is usually computed from spikein RNA and ensures that the simulated doublets have the correct ratio of contributions from the original cells.
It is possible to set both of these arguments as they are orthogonal to each other.
Setting size.factors.content
will not affect the calculation of lognormalized expression values from x
.
Conversely, setting size.factors.norm
will not affect the ratio in which cells are added together when simulating doublets.
If force.match=TRUE
, simulated doublets will be remapped to the nearest neighbours in the original data.
This is done by taking the (tricubeweighted) average of the PC scores for the force.k
nearest neighbors.
The tricube bandwidth for remapping is chosen by taking the median distance and multiplying it by force.ndist
, to protect against later neighbours that might be outliers.
The aim is to adjust for unknown differences in RNA content that would cause the simulated doublets to be systematically displaced from their true locations.
However, it may also result in spuriously high scores for single cells that happen to be close to a cluster of simulated doublets.
A numeric vector of doublet scores for each cell in x
.
Aaron Lun
Lun ATL (2018). Detecting doublet cells with scran. https://ltla.github.io/SingleCellThoughts/software/doublet_detection/bycell.html
doubletCluster
, to detect doublet clusters.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20  # Mocking up an example.
set.seed(100)
ngenes < 1000
mu1 < 2^rnorm(ngenes)
mu2 < 2^rnorm(ngenes)
mu3 < 2^rnorm(ngenes)
mu4 < 2^rnorm(ngenes)
counts.1 < matrix(rpois(ngenes*100, mu1), nrow=ngenes) # Pure type 1
counts.2 < matrix(rpois(ngenes*100, mu2), nrow=ngenes) # Pure type 2
counts.3 < matrix(rpois(ngenes*100, mu3), nrow=ngenes) # Pure type 3
counts.4 < matrix(rpois(ngenes*100, mu4), nrow=ngenes) # Pure type 4
counts.m < matrix(rpois(ngenes*20, mu1+mu2), nrow=ngenes) # Doublets (1 & 2)
counts < cbind(counts.1, counts.2, counts.3, counts.4, counts.m)
clusters < rep(1:5, c(rep(100, 4), ncol(counts.m)))
# Find potential doublets.
scores < doubletCells(counts)
boxplot(split(log10(scores), clusters))

Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.