tweeDEseq (version 1.18.0)

filterCounts: Count data filtering

Description

Filter count data to remove lowly expressed genes.

Usage

filterCounts(counts, cpm.cutoff=0.5, n.samples.cutoff=2, mean.cpm.cutoff=0, lib.sizes=NULL)

Arguments

counts

numeric data.frame or matrix containing the count data.

cpm.cutoff

expression level cutoff defined as the minimum number of counts per million. By default this is set to 0.5 counts per million.

n.samples.cutoff

minimum number of samples where a gene should meet the counts per million cutoff (cpm.cutoff) in order to be kept as part of the count data matrix. When n.samples.cutoff is a number between 0 and 1, then it is interpreted as the fraction of samples that should meet the counts per million cutoff (cpm.cutoff).

mean.cpm.cutoff

minimum mean of counts per million cutoff that a gene should meet in order to be kept. When the value of this argument is larger than 0 then it overrules the other arguments cpm.cutoff and n.samples.cutoff.

lib.sizes

vector of the total number of reads to be considered per sample/library. If lib.sizes=NULL (default) then these quantities are estimated as the column sums in the input matrix of counts.

Value

A matrix of filtered genes.

Details

This function removes genes with very low expression level defined in terms of a minimum number of counts per million occurring in a minimum number of samples. Such a policy was described by Davis McCarthy in a message at the bioc-sig-sequencing mailing list. By default, this function keeps genes that are expressed at a level of 0.5 counts per million or greater in at least two samples. Alternatively, one can use the mean.cpm.cutoff to set a minimum mean expression level through all the samples.

References

Davis McCarthy, https://stat.ethz.ch/pipermail/bioc-sig-sequencing/2011-June/002072.html.

See Also

normalizeCounts

Examples

# Generate a random matrix of counts
counts <- matrix(rPT(n=1000, a=0.5, mu=10, D=5), ncol = 40)

dim(counts)

# Filter genes with requiring the minimum expression level on every sample
filteredCounts <- filterCounts(counts, n.samples.cutoff=dim(counts)[2])

dim(filteredCounts)