detectLowQualityCells: detectLowQualityCells

In martinenge/RNAseqFunctions: RNA sequencing functions for the Enge lab

Description

It is often the case that some samples from sequencing experiments are of low quality, in many cases due to issues during the sample preperation stage. Due to the fact that these samples represent a high level of technical noise, it is often desirable to remove these before downstream analysis which is facilitated by this function. The function achieves this using two methods. First, the mincount argument detects samples whose sum across all genes is > mincount. Second, we utilize a house keeping gene and assume its expression to be normally distributed. We then detect samples where the probability of the expression for the house keeping gene in that sample is greater than the quantile.cut argument.

Arguments

counts	data.frame; A data frame with counts data with gene names as rownames and sample names as colnames.
mincount	numeric; A minimum colSum for which columns with a higher colSum will be detected. Default = 4e5.
geneName	character; The gene name to use for the quantile cutoff. This must be present in the rownames of the counts argument. Default is ACTB.
quantileCut	numeric; This indicates probability at which the quantile cutoff will be calculated using the normal distribution. Default = 0.01.

Value

A logical vector with length = ncol(counts) that is TRUE when the counts data.frame column contains a sample with colSums > mincount.

Author(s)

Jason Serviss

Examples

c <- moveGenesToRownames(testingCounts)[1:12, ]
detectLowQualityCells(c, geneName = "ACTB", mincount = 30)

martinenge/RNAseqFunctions documentation built on May 28, 2019, 3:10 p.m.