HolmBonferroni: Apply Holm-Bonferroni method to adjust for multiple tests

Description

Apply the Holm-Bonferroni method to adjust for multiple hypothesis tests performed on quartets from a data set of gene trees.

Usage

HolmBonferroni(pTable, model, alpha = 0.05)

Value

the same table, with rows reordered, and 2 new columns of 1) adjusted p-values, and 2) "Y" or "N" for indicating "reject" or "fail to reject"

Arguments

pTable: a table of quartets with p-values, as computed by quartetTreeTestInd or quartetStarTestInd
model: one of "T1", "T3", or "star", where pTable contains a column p_model of p-values
alpha: test level, for rejection of adjusted p-values less than or equal to alpha

Details

When p-values are computed for each quartet using quartetTreeTestInd or quartetStarTestInd, multiple comparisons are being done for one dataset. The Holm-Bonferroni method HolmBonf79MSCquartets adjusts these p-values upward, controlling the familywise error rate. The probability of at least one false discovery (rejection of the null hypothesis) is no more than the significance level.

The Holm-Bonferroni method does not require that test hypotheses are independent, which is important for its application to quartet counts presumed to have arisen on a single species tree.

This can be a low power test (often failing to reject when the null hypothesis is false). In particular for the T1 and T3 tests, it does not consider the relationships between edge lengths for different sets of four taxa.

Warning: Output of this function should not be used as input for other MSCquartets functions, as it reorders the quartets in the table.

References

HolmBonf79MSCquartets

Examples

Run this code

gtrees=read.tree(file=system.file("extdata","dataGeneTreeSample",package="MSCquartets"))
taxanames=taxonNames(gtrees)
QT=quartetTable(gtrees,taxanames[1:6])
RQT=quartetTableResolved(QT)
pTable=quartetTreeTestInd(RQT,"T3")
pTable[1:10,]
HBpTable=HolmBonferroni(pTable,"T3",.05)
HBpTable[1:10,]

Run the code above in your browser using DataLab

Last chance! 50% off unlimited learning