average.pval.replicates-methods: average.pval.replicates

In drmjc/microarrays: A collection of microarray and genomic helper code.

Description

average.pval.replicates

Arguments

x	a matrix or data.frame of p-values in [0,1]
classes	a character, numeric or factor vector of classes that each sample belongs to. in this instance, class usually represents the original sample from which each array has been generated from.

Details

Given a 2D table of pvalues, such as the detection pvalue for each probe for each sample, compute an average p-value for those samples which are replicates of each other.
This uses the more correct calculation of p*(1-ln(p)), where p = p1*p2, which is easily generalised to >2 replicates.

Value

a matrix or data.frame of 'averaged' p.values. See Details.

Note

Since detection pvals=0.00000 are quite common, and log(0) == NaN, if any of the p-values are 0, then the averaged p-value is also set to 0. I'm not 100% sure this is the best approach – it's certainly correct of all values are 0, but theoretically, 1 sample could have P=1.0 and the other P=0.

Author(s)

Mark Cowley, Mark Pinese

drmjc/microarrays documentation built on May 15, 2019, 2:26 p.m.