# plotCellTypePropsMeanVar: Plot cell type proportions versus variances In Oshlack/speckle: Statistical methods for analysing single cell RNA-seq data

 plotCellTypePropsMeanVar R Documentation

## Plot cell type proportions versus variances

### Description

This function returns a plot of the log10(proportion) versus log10(variance) given a matrix of cell type counts. The rows are the clusters/cell types and the columns are the samples.

### Usage

```plotCellTypePropsMeanVar(x)
```

### Arguments

 `x` a matrix or table of counts

### Details

The expected variance under a binomial distribution is shown in the solid line, and the points represent the observed variance for each cell type/row in the counts table. The blue line shows the empirical Bayes variance that is used in `propeller`.

The mean and variance for each cell type is calculated across all samples.

a base R plot

Belinda Phipson

### Examples

```library(limma)
# Generate some data
# Total number of samples
nsamp <- 10
# True cell type proportions
p <- c(0.05, 0.15, 0.35, 0.45)

# Parameters for beta distribution
a <- 40
b <- a*(1-p)/p

# Sample total cell counts per sample from negative binomial distribution
numcells <- rnbinom(nsamp,size=20,mu=5000)
true.p <- matrix(c(rbeta(nsamp,a,b[1]),rbeta(nsamp,a,b[2]),
rbeta(nsamp,a,b[3]),rbeta(nsamp,a,b[4])),byrow=TRUE, ncol=nsamp)

counts <- matrix(NA,ncol=nsamp, nrow=nrow(true.p))
rownames(counts) <- paste("c",0:(nrow(true.p)-1), sep="")
for(j in 1:length(p)){
counts[j,] <- rbinom(nsamp, size=numcells, prob=true.p[j,])
}

plotCellTypePropsMeanVar(counts)

```

Oshlack/speckle documentation built on Oct. 16, 2022, 9:39 a.m.