DESeq2_tidiers: Tidying methods for DESeq2 DESeqDataSet objects

In StoreyLab/biobroom: Turn Bioconductor objects into tidy data frames

Description

This reshapes a DESeq2 expressionset object into a tidy format. If the dataset contains hypothesis test results (p-values and estimates), this summarizes one row per gene per possible contrast.

Usage

## S3 method for class 'DESeqDataSet'
tidy(x, colData = FALSE, intercept = FALSE, ...)

## S3 method for class 'DESeqResults'
tidy(x, ...)

Arguments

x	DESeqDataSet object
colData	whether colData should be included in the tidied output for those in the DESeqDataSet object. If dataset includes hypothesis test results, this is ignored
intercept	whether to include hypothesis test results from the (Intercept) term. If dataset does not include hypothesis testing, this is ignored
...	extra arguments (not used)

Details

colDat=TRUE adds covariates from colData to the data frame.

Value

If the dataset contains results (p-values and log2 fold changes), the result is a data frame with the columns

term	The contrast being tested, as given to results
gene	gene ID
baseMean	mean abundance level
estimate	estimated log2 fold change
stderror	standard error in log2 fold change estimate
statistic	test statistic
p.value	p-value
p.adjusted	adjusted p-value

If the dataset does not contain results (DESeq has not been run on it), tidy defaults to tidying the counts in the dataset:

gene	gene ID
sample	sample ID
count	number of reads in this gene in this sample

If colData = TRUE, it also merges this with the columns present in colData(x).

Examples

# From DESeq2 documentation

if (require("DESeq2")) {
    dds <- makeExampleDESeqDataSet(betaSD = 1)

    tidy(dds)
    # With design included
    tidy(dds, colData=TRUE)

    # add a noise confounding effect
    colData(dds)$noise <- rnorm(nrow(colData(dds)))
    design(dds) <- (~ condition + noise)

    # perform differential expression tests
    ddsres <- DESeq(dds, test = "Wald")
    # now results are per-gene, per-term
    tidied <- tidy(ddsres)
    tidied

    if (require("ggplot2")) {
        ggplot(tidied, aes(p.value)) + geom_histogram(binwidth = .05) +
            facet_wrap(~ term, scale = "free_y")
    }
}

StoreyLab/biobroom documentation built on May 9, 2019, 3:09 p.m.