HTqPCR (version 1.26.0)

mannwhitneyCtData: Differentially expressed features with qPCR: Mann-Whitney

Description

Function for calculating p-values across two groups for the features present in high-throughput qPCR data, such as from TaqMan Low Density Arrays. Also known as two sample Wilcoxon test.

Usage

mannwhitneyCtData(q, groups = NULL, calibrator, alternative = "two.sided", paired = FALSE, replicates = TRUE, sort = TRUE, stringent = TRUE, p.adjust = "BH", ...)

Arguments

q

qPCRset object.

groups

factor, assigning each sample to one of two groups.

calibrator

which of the two groups is to be considered as the reference and not the test? Defaults to the first group in groups.

alternative

character string (first letter is enough), specifying the alternative hypothesis, "two.sided" (default), "greater" or "less".

paired

logical, should a paired t-test be used.

replicates

logical, if replicated genes are present on the array, the statistics will be calculated for all the replicates combined, rather than the individual wells.

sort

boolean, should the output be sorted by p-values.

stringent

boolean, for flagging results as "Undetermined". See details.

p.adjust

character string, which method to use for p-value adjustment for multiple testing. See details.

...

any other arguments will be passed to the wilcox.test function.

Value

genes

The names of the features on the card.

feature.pos

The featurePos of the genes. If replicated genes are used, the feature positions will be concatenated together.

MB.test

The name and value of the test statistic.

p.value

The corresponding p-value.

ddCt

The delta delta Ct values.

FC

The fold change; 2^(-ddCt).

meanCalibrator

The average expression level of each gene in the calibrator sample(s).

meanTarget

The average expression level of each gene in the target sample(s).

categoryCalibrator

The category of the Ct values ("OK", "Undetermined") across the calibrator.

categoryTarget

Ditto for the target.

Details

Once the Ct values have been normalised, differential expression can be calculated. This function deals with just the simple case, where there are two types of samples to compare. For a parametric test see ttestCtData and limmaCtData for more complex studies. The underlying statistics is calculated by wilcox.test. Due to the high possibility of ties for each feature between samples, the test is run with exact=FALSE.

All results are assigned to a category, either "OK" or "Undetermined" depending on the input Ct values. If stringent=TRUE any unreliable or undetermined measurements among technical and biological replicates will result in the final result being "Undetermined". For stringent=FALSE the result will be "OK" unless at least half of the Ct values for a given gene are unreliable/undetermined.

The argument p.adjust is passed on to the p.adjust function. Options include e.g. "BH" (Benjamini & Hochberg, the default), "fdr" and "bonferroni". See p.adjust for more information on the individual methods.

See Also

wilcox.test, ttestCtData, limmaCtData. plotCtRQ and plotCtSignificance can be used for visualising the results.

Examples

# Load example preprocessed data
data(qPCRpros)
# Test between two groups, collapsing replicated features
diff.exp <- mannwhitneyCtData(qPCRpros[,1:4], groups=factor(c("A", "B", "B", "A")), calibrator="B")
diff.exp[1:10,]
# The same test, taking replicated features individually
diff.exp <- mannwhitneyCtData(qPCRpros[,1:4], groups=factor(c("A", "B", "B", "A")), calibrator="B", replicates=FALSE)
# Using another method for p-value adjustment
diff.exp <- mannwhitneyCtData(qPCRpros[,1:4], groups=factor(c("A", "B", "B", "A")), calibrator="B", p.adjust="holm")