DGEobj.utils (version 1.0.6)

runPower: Run a power analysis on counts and design matrix

Description

Take a counts matrix and design matrix and return a power analysis using the RNASeqPower package. The counts matrix should be pre-filtered to remove non-expressed genes using an appropriate filtering criteria. The design matrix should describe the major sources of variation so the procedure can dial out those known effects for the power calculations.

Usage

runPower(
  countsMatrix,
  designMatrix,
  depth = c(10, 100, 1000),
  N = c(3, 6, 10, 20),
  FDR = c(0.05, 0.1),
  effectSize = c(1.2, 1.5, 2),
  includePlots = FALSE
)

Arguments

countsMatrix

A counts matrix or dataframe of numeric data. (Required)

designMatrix

A design matrix or dataframe of numeric data. (Required)

depth

A set of depth to use in the calculations. The default depths of c(10, 100, 1000) respectively represent a detection limit, below average expression, and median expression levels, expressed in read count units.

N

A set of N value to report power for. (Default = c(3, 6, 10, 20))

FDR

FDR thresholds to filter for for FDR vs. Power graph. (Default = c(0.05, 0.1))

effectSize

A set of fold change values to test. (Default = c(1.2, 1.5, 2))

includePlots

controls adding tow plots to the returned dataframe (Default = FALSE). The two plots are; a ROC curve (FDR vs. Power) and a plot of N vs. Power. Possible values to pass:

• FALSE or NULL: Disable plots

• TRUE or "canvasXpress": returns "canvasXpress" plots.

• "ggplot": returns "ggplot" plots.

Value

a dataframe of power calculations or a list of the dataframe and defined plots as defined by the "includePlots" argument.

Details

Note, both 'RNASeqPower' and 'statmod' packages are required to run this function as follow:

• 'RNASeqPower' package is required to run power analysis on the given counts matrix and design matrix.

• 'statmod' package is required to run estimate dispersion calculations

If includePlots = FALSE (the default) or NULL, the function will return a tall skinny dataframe of power calculations for various requested combinations of N and significance thresholds.

If includePlots = TRUE, "canvasXpress" or "ggplot", a list is returned with an additional two "canvasXpress" or ggplots (plots) to the dataframe.

Examples

# NOT RUN {
    # NOTE: Requires the RNASeqPower, statmod, and edgeR packages

    dgeObj <- readRDS(system.file("exampleObj.RDS", package = "DGEobj"))
    counts <- dgeObj$counts
    dm     <- DGEobj::getType(dgeObj, type = "designMatrix")[[1]]

    resultList <- runPower(countsMatrix = counts,
                           designMatrix = dm,
                           includePlots = TRUE)

    head(resultList[[1]]) # dataframe
    resultList[[2]]       # ROC Curves Plot
    resultList[[3]]       # N vs Power Plot
# }
# NOT RUN {
# }